Datasets:

COLING

/

COLING2024-papers

like 0

Follow

International Conference on Computational Linguistics 1

https://aclanthology.org/2024.lrec-main.701.bib

https://aclanthology.org/2024.lrec-main.701/

@inproceedings{zanotto-etal-2024-grit, title = "{GRIT}: A Dataset of Group Reference Recognition in {I}talian", author = "Zanotto, Sergio E. and Yu, Qi and Butt, Miriam and Frassinelli, Diego", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.701", pages = "7963--7970", abstract = "For the analysis of political discourse a reliable identification of group references, i.e., linguistic components that refer to individuals or groups of people, is useful. However, the task of automatically recognizing group references has not yet gained much attention within NLP. To address this gap, we introduce GRIT (Group Reference for Italian), a large-scale, multi-domain manually annotated dataset for group reference recognition in Italian. GRIT represents a new resource for automatic and generalizable recognition of group references. With this dataset, we aim to establish group reference recognition as a valid classification task, which extends the domain of Named Entity Recognition by expanding its focus to literal and figurative mentions of social groups. We verify the potential of achieving automated group reference recognition for Italian through an experiment employing a fine-tuned BERT model. Our experimental results substantiate the validity of the task, implying a huge potential for applying automated systems to multiple fields of analysis, such as political text or social media analysis.", }

For the analysis of political discourse a reliable identification of group references, i.e., linguistic components that refer to individuals or groups of people, is useful. However, the task of automatically recognizing group references has not yet gained much attention within NLP. To address this gap, we introduce GRIT (Group Reference for Italian), a large-scale, multi-domain manually annotated dataset for group reference recognition in Italian. GRIT represents a new resource for automatic and generalizable recognition of group references. With this dataset, we aim to establish group reference recognition as a valid classification task, which extends the domain of Named Entity Recognition by expanding its focus to literal and figurative mentions of social groups. We verify the potential of achieving automated group reference recognition for Italian through an experiment employing a fine-tuned BERT model. Our experimental results substantiate the validity of the task, implying a huge potential for applying automated systems to multiple fields of analysis, such as political text or social media analysis.

GRIT: A Dataset of Group Reference Recognition in Italian

lrec-main.701

Poster

https://aclanthology.org/2024.lrec-main.702.bib

https://aclanthology.org/2024.lrec-main.702/

@inproceedings{ren-etal-2024-grounded, title = "Grounded Multimodal Procedural Entity Recognition for Procedural Documents: A New Dataset and Baseline", author = "Ren, Haopeng and Zeng, Yushi and Cai, Yi and Ye, Zhenqi and Yuan, Li and Zhu, Pinli", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.702", pages = "7971--7981", abstract = "Much of commonsense knowledge in real world is the form of procudures or sequences of steps to achieve particular goals. In recent years, knowledge extraction on procedural documents has attracted considerable attention. However, they often focus on procedural text but ignore a common multimodal scenario in the real world. Images and text can complement each other semantically, alleviating the semantic ambiguity suffered in text-only modality. Motivated by these, in this paper, we explore a problem of grounded multimodal procedural entity recognition (GMPER), aiming to detect the entity and the corresponding bounding box groundings in image (i.e., visual entities). A new dataset (Wiki-GMPER) is bult and extensive experiments are conducted to evaluate the effectiveness of our proposed model.", }

Much of commonsense knowledge in real world is the form of procudures or sequences of steps to achieve particular goals. In recent years, knowledge extraction on procedural documents has attracted considerable attention. However, they often focus on procedural text but ignore a common multimodal scenario in the real world. Images and text can complement each other semantically, alleviating the semantic ambiguity suffered in text-only modality. Motivated by these, in this paper, we explore a problem of grounded multimodal procedural entity recognition (GMPER), aiming to detect the entity and the corresponding bounding box groundings in image (i.e., visual entities). A new dataset (Wiki-GMPER) is bult and extensive experiments are conducted to evaluate the effectiveness of our proposed model.

Grounded Multimodal Procedural Entity Recognition for Procedural Documents: A New Dataset and Baseline

lrec-main.702

Poster

https://aclanthology.org/2024.lrec-main.703.bib

https://aclanthology.org/2024.lrec-main.703/

@inproceedings{plum-etal-2024-guided, title = "Guided Distant Supervision for Multilingual Relation Extraction Data: Adapting to a New Language", author = "Plum, Alistair and Ranasinghe, Tharindu and Purschke, Christoph", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.703", pages = "7982--7992", abstract = "Relation extraction is essential for extracting and understanding biographical information in the context of digital humanities and related subjects. There is a growing interest in the community to build datasets capable of training machine learning models to extract relationships. However, annotating such datasets can be expensive and time-consuming, in addition to being limited to English. This paper applies guided distant supervision to create a large biographical relationship extraction dataset for German. Our dataset, composed of more than 80,000 instances for nine relationship types, is the largest biographical German relationship extraction dataset. We also create a manually annotated dataset with 2000 instances to evaluate the models and release it together with the dataset compiled using guided distant supervision. We train several state-of-the-art machine learning models on the automatically created dataset and release them as well. Furthermore, we experiment with multilingual and cross-lingual zero-shot experiments that could benefit many low-resource languages.", }

Relation extraction is essential for extracting and understanding biographical information in the context of digital humanities and related subjects. There is a growing interest in the community to build datasets capable of training machine learning models to extract relationships. However, annotating such datasets can be expensive and time-consuming, in addition to being limited to English. This paper applies guided distant supervision to create a large biographical relationship extraction dataset for German. Our dataset, composed of more than 80,000 instances for nine relationship types, is the largest biographical German relationship extraction dataset. We also create a manually annotated dataset with 2000 instances to evaluate the models and release it together with the dataset compiled using guided distant supervision. We train several state-of-the-art machine learning models on the automatically created dataset and release them as well. Furthermore, we experiment with multilingual and cross-lingual zero-shot experiments that could benefit many low-resource languages.

Guided Distant Supervision for Multilingual Relation Extraction Data: Adapting to a New Language

lrec-main.703

Poster

2403.17143

https://aclanthology.org/2024.lrec-main.704.bib

https://aclanthology.org/2024.lrec-main.704/

@inproceedings{son-etal-2024-hae, title = "{HAE}-{RAE} Bench: Evaluation of {K}orean Knowledge in Language Models", author = "Son, Guijin and Lee, Hanwool and Kim, Suwan and Kim, Huiseo and Lee, Jae cheol and Yeom, Je Won and Jung, Jihyu and Kim, Jung woo and Kim, Songseong", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.704", pages = "7993--8007", abstract = "Large language models (LLMs) trained on massive corpora demonstrate impressive capabilities in a wide range of tasks. While there are ongoing efforts to adapt these models to languages beyond English, the attention given to their evaluation methodologies remains limited. Current multilingual benchmarks often rely on back translations or re-implementations of English tests, limiting their capacity to capture unique cultural and linguistic nuances. To bridge this gap for the Korean language, we introduce the HAE-RAE Bench, a dataset curated to challenge models lacking Korean cultural and contextual depth. The dataset encompasses six downstream tasks across four domains: vocabulary, history, general knowledge, and reading comprehension. Unlike traditional evaluation suites focused on token and sequence classification or mathematical and logical reasoning, the HAE-RAE Bench emphasizes a model{'}s aptitude for recalling Korean-specific knowledge and cultural contexts. Comparative analysis with prior Korean benchmarks indicates that the HAE-RAE Bench presents a greater challenge to non-Korean models by disturbing abilities and knowledge learned from English being transferred.", }

Large language models (LLMs) trained on massive corpora demonstrate impressive capabilities in a wide range of tasks. While there are ongoing efforts to adapt these models to languages beyond English, the attention given to their evaluation methodologies remains limited. Current multilingual benchmarks often rely on back translations or re-implementations of English tests, limiting their capacity to capture unique cultural and linguistic nuances. To bridge this gap for the Korean language, we introduce the HAE-RAE Bench, a dataset curated to challenge models lacking Korean cultural and contextual depth. The dataset encompasses six downstream tasks across four domains: vocabulary, history, general knowledge, and reading comprehension. Unlike traditional evaluation suites focused on token and sequence classification or mathematical and logical reasoning, the HAE-RAE Bench emphasizes a model{'}s aptitude for recalling Korean-specific knowledge and cultural contexts. Comparative analysis with prior Korean benchmarks indicates that the HAE-RAE Bench presents a greater challenge to non-Korean models by disturbing abilities and knowledge learned from English being transferred.

HAE-RAE Bench: Evaluation of Korean Knowledge in Language Models

lrec-main.704

Poster

2309.02706

https://aclanthology.org/2024.lrec-main.705.bib

https://aclanthology.org/2024.lrec-main.705/

@inproceedings{mubarak-etal-2024-halwasa, title = "Halwasa: Quantify and Analyze Hallucinations in Large Language Models: {A}rabic as a Case Study", author = "Mubarak, Hamdy and Al-Khalifa, Hend and Alkhalefah, Khaloud Suliman", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.705", pages = "8008--8015", abstract = "Large Language Models (LLMs) have shown superb abilities to generate texts that are indistinguishable from human-generated texts in many cases. However, sometimes they generate false, incorrect, or misleading content, which is often described as {``}hallucinations{''}. Quantifying and analyzing hallucination in LLMs can increase their reliability and usage. While hallucination is being actively studied for English and other languages, and different benchmarking datsets have been created, this area is not studied at all for Arabic. In our paper, we create the first Arabic dataset that contains 10K of generated sentences by LLMs and annotate it for factuality and correctness. We provide detailed analysis of the dataset to analyze factual and linguistic errors. We found that 25{\%} of the generated sentences are factually incorrect. We share the dataset with the research community.", }

Large Language Models (LLMs) have shown superb abilities to generate texts that are indistinguishable from human-generated texts in many cases. However, sometimes they generate false, incorrect, or misleading content, which is often described as {``}hallucinations{''}. Quantifying and analyzing hallucination in LLMs can increase their reliability and usage. While hallucination is being actively studied for English and other languages, and different benchmarking datsets have been created, this area is not studied at all for Arabic. In our paper, we create the first Arabic dataset that contains 10K of generated sentences by LLMs and annotate it for factuality and correctness. We provide detailed analysis of the dataset to analyze factual and linguistic errors. We found that 25{\%} of the generated sentences are factually incorrect. We share the dataset with the research community.

Halwasa: Quantify and Analyze Hallucinations in Large Language Models: Arabic as a Case Study

lrec-main.705

Poster

https://aclanthology.org/2024.lrec-main.706.bib

https://aclanthology.org/2024.lrec-main.706/

@inproceedings{kumar-etal-2024-harmpot, title = "{H}arm{P}ot: An Annotation Framework for Evaluating Offline Harm Potential of Social Media Text", author = "Kumar, Ritesh and Bhalla, Ojaswee and Vanthi, Madhu and Wani, Shehlat Maknoon and Singh, Siddharth", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.706", pages = "8016--8034", abstract = "In this paper, we discuss the development of an annotation schema to build datasets for evaluating the offline harm potential of social media texts. We define {``}harm potential{''} as the potential for an online public post to cause real-world physical harm (i.e., violence). Understanding that real-world violence is often spurred by a web of triggers, often combining several online tactics and pre-existing intersectional fissures in the social milieu, to result in targeted physical violence, we do not focus on any single divisive aspect (i.e., caste, gender, religion, or other identities of the victim and perpetrators) nor do we focus on just hate speech or mis/dis-information. Rather, our understanding of the intersectional causes of such triggers focuses our attempt at measuring the harm potential of online content, irrespective of whether it is hateful or not. In this paper, we discuss the development of a framework/annotation schema that allows annotating the data with different aspects of the text including its socio-political grounding and intent of the speaker (as expressed through mood and modality) that together contribute to it being a trigger for offline harm. We also give a comparative analysis and mapping of our framework with some of the existing frameworks.", }

In this paper, we discuss the development of an annotation schema to build datasets for evaluating the offline harm potential of social media texts. We define {``}harm potential{''} as the potential for an online public post to cause real-world physical harm (i.e., violence). Understanding that real-world violence is often spurred by a web of triggers, often combining several online tactics and pre-existing intersectional fissures in the social milieu, to result in targeted physical violence, we do not focus on any single divisive aspect (i.e., caste, gender, religion, or other identities of the victim and perpetrators) nor do we focus on just hate speech or mis/dis-information. Rather, our understanding of the intersectional causes of such triggers focuses our attempt at measuring the harm potential of online content, irrespective of whether it is hateful or not. In this paper, we discuss the development of a framework/annotation schema that allows annotating the data with different aspects of the text including its socio-political grounding and intent of the speaker (as expressed through mood and modality) that together contribute to it being a trigger for offline harm. We also give a comparative analysis and mapping of our framework with some of the existing frameworks.

HarmPot: An Annotation Framework for Evaluating Offline Harm Potential of Social Media Text

lrec-main.706

Poster

2403.11108

https://aclanthology.org/2024.lrec-main.707.bib

https://aclanthology.org/2024.lrec-main.707/

@inproceedings{qian-etal-2024-harnessing, title = "Harnessing the Power of Large Language Model for Uncertainty Aware Graph Processing", author = "Qian, Zhenyu and Qian, Yiming and Song, Yuting and Gao, Fei and Jin, Hai and Yu, Chen and Xie, Xia", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.707", pages = "8035--8049", abstract = "Handling graph data is one of the most difficult tasks. Traditional techniques, such as those based on geometry and matrix factorization, rely on assumptions about the data relations that become inadequate when handling large and complex graph data. On the other hand, deep learning approaches demonstrate promising results in handling large graph data, but they often fall short of providing interpretable explanations. To equip the graph processing with both high accuracy and explainability, we introduce a novel approach that harnesses the power of a large language model (LLM), enhanced by an uncertainty-aware module to provide a confidence score on the generated answer. We experiment with our approach on two graph processing tasks: few-shot knowledge graph completion and graph classification. Our results demonstrate that through parameter efficient fine-tuning, the LLM surpasses state-of-the-art algorithms by a substantial margin across ten diverse benchmark datasets. Moreover, to address the challenge of explainability, we propose an uncertainty estimation based on perturbation, along with a calibration scheme to quantify the confidence scores of the generated answers. Our confidence measure achieves an AUC of 0.8 or higher on seven out of the ten datasets in predicting the correctness of the answer generated by LLM.", }

Handling graph data is one of the most difficult tasks. Traditional techniques, such as those based on geometry and matrix factorization, rely on assumptions about the data relations that become inadequate when handling large and complex graph data. On the other hand, deep learning approaches demonstrate promising results in handling large graph data, but they often fall short of providing interpretable explanations. To equip the graph processing with both high accuracy and explainability, we introduce a novel approach that harnesses the power of a large language model (LLM), enhanced by an uncertainty-aware module to provide a confidence score on the generated answer. We experiment with our approach on two graph processing tasks: few-shot knowledge graph completion and graph classification. Our results demonstrate that through parameter efficient fine-tuning, the LLM surpasses state-of-the-art algorithms by a substantial margin across ten diverse benchmark datasets. Moreover, to address the challenge of explainability, we propose an uncertainty estimation based on perturbation, along with a calibration scheme to quantify the confidence scores of the generated answers. Our confidence measure achieves an AUC of 0.8 or higher on seven out of the ten datasets in predicting the correctness of the answer generated by LLM.

Harnessing the Power of Large Language Model for Uncertainty Aware Graph Processing

lrec-main.707

Poster

2404.00589

https://aclanthology.org/2024.lrec-main.708.bib

https://aclanthology.org/2024.lrec-main.708/

@inproceedings{ignat-etal-2024-solved, title = "Has It All Been Solved? Open {NLP} Research Questions Not Solved by Large Language Models", author = "Ignat, Oana and Jin, Zhijing and Abzaliev, Artem and Biester, Laura and Castro, Santiago and Deng, Naihao and Gao, Xinyi and Gunal, Aylin Ece and He, Jacky and Kazemi, Ashkan and Khalifa, Muhammad and Koh, Namho and Lee, Andrew and Liu, Siyang and Min, Do June and Mori, Shinka and Nwatu, Joan C. and Perez-Rosas, Veronica and Shen, Siqi and Wang, Zekun and Wu, Winston and Mihalcea, Rada", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.708", pages = "8050--8094", abstract = "Recent progress in large language models (LLMs) has enabled the deployment of many generative NLP applications. At the same time, it has also led to a misleading public discourse that {``}it{'}s all been solved.{''} Not surprisingly, this has, in turn, made many NLP researchers {--} especially those at the beginning of their careers {--} worry about what NLP research area they should focus on. Has it all been solved, or what remaining questions can we work on regardless of LLMs? To address this question, this paper compiles NLP research directions rich for exploration. We identify fourteen different research areas encompassing 45 research directions that require new research and are not directly solvable by LLMs. While we identify many research areas, many others exist; we do not cover areas currently addressed by LLMs, but where LLMs lag behind in performance or those focused on LLM development. We welcome suggestions for other research directions to include: https://bit.ly/nlp-era-llm.", }

Recent progress in large language models (LLMs) has enabled the deployment of many generative NLP applications. At the same time, it has also led to a misleading public discourse that {``}it{'}s all been solved.{''} Not surprisingly, this has, in turn, made many NLP researchers {--} especially those at the beginning of their careers {--} worry about what NLP research area they should focus on. Has it all been solved, or what remaining questions can we work on regardless of LLMs? To address this question, this paper compiles NLP research directions rich for exploration. We identify fourteen different research areas encompassing 45 research directions that require new research and are not directly solvable by LLMs. While we identify many research areas, many others exist; we do not cover areas currently addressed by LLMs, but where LLMs lag behind in performance or those focused on LLM development. We welcome suggestions for other research directions to include: https://bit.ly/nlp-era-llm.

Has It All Been Solved? Open NLP Research Questions Not Solved by Large Language Models

lrec-main.708

Poster

2305.12544

https://huggingface.co/papers/2305.12544

https://aclanthology.org/2024.lrec-main.709.bib

https://aclanthology.org/2024.lrec-main.709/

@inproceedings{vladika-etal-2024-healthfc, title = "{H}ealth{FC}: Verifying Health Claims with Evidence-Based Medical Fact-Checking", author = "Vladika, Juraj and Schneider, Phillip and Matthes, Florian", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.709", pages = "8095--8107", abstract = "In the digital age, seeking health advice on the Internet has become a common practice. At the same time, determining the trustworthiness of online medical content is increasingly challenging. Fact-checking has emerged as an approach to assess the veracity of factual claims using evidence from credible knowledge sources. To help advance automated Natural Language Processing (NLP) solutions for this task, in this paper we introduce a novel dataset HealthFC. It consists of 750 health-related claims in German and English, labeled for veracity by medical experts and backed with evidence from systematic reviews and clinical trials. We provide an analysis of the dataset, highlighting its characteristics and challenges. The dataset can be used for NLP tasks related to automated fact-checking, such as evidence retrieval, claim verification, or explanation generation. For testing purposes, we provide baseline systems based on different approaches, examine their performance, and discuss the findings. We show that the dataset is a challenging test bed with a high potential for future use.", }

In the digital age, seeking health advice on the Internet has become a common practice. At the same time, determining the trustworthiness of online medical content is increasingly challenging. Fact-checking has emerged as an approach to assess the veracity of factual claims using evidence from credible knowledge sources. To help advance automated Natural Language Processing (NLP) solutions for this task, in this paper we introduce a novel dataset HealthFC. It consists of 750 health-related claims in German and English, labeled for veracity by medical experts and backed with evidence from systematic reviews and clinical trials. We provide an analysis of the dataset, highlighting its characteristics and challenges. The dataset can be used for NLP tasks related to automated fact-checking, such as evidence retrieval, claim verification, or explanation generation. For testing purposes, we provide baseline systems based on different approaches, examine their performance, and discuss the findings. We show that the dataset is a challenging test bed with a high potential for future use.

HealthFC: Verifying Health Claims with Evidence-Based Medical Fact-Checking

lrec-main.709

Poster

2309.08503

https://aclanthology.org/2024.lrec-main.710.bib

https://aclanthology.org/2024.lrec-main.710/

@inproceedings{lee-etal-2024-hierarchical, title = "Hierarchical Graph Convolutional Network Approach for Detecting Low-Quality Documents", author = "Lee, Jaeyoung and Jang, Joonwon and Kim, Misuk", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.710", pages = "8108--8121", abstract = "Consistency within a document is a crucial feature indicative of its quality. Recently, within the vast amount of information produced across various media, there exists a significant number of low-quality documents that either lack internal consistency or contain content utterly unrelated to their headlines. Such low-quality documents induce fatigue in readers and undermine the credibility of the media source that provided them. Consequently, research to automatically detect these low-quality documents based on natural language processing is imperative. In this study, we introduce a hierarchical graph convolutional network (HGCN) that can detect internal inconsistencies within a document and incongruences between the title and body. Moreover, we constructed the Inconsistency Dataset, leveraging published news data and its meta-data, to train our model to detect document inconsistencies. Experimental results demonstrated that the HGCN achieved superior performance with an accuracy of 91.20{\%} on our constructed Inconsistency Dataset, outperforming other comparative models. Additionally, on the publicly available incongruent-related dataset, the proposed methodology demonstrated a performance of 92.00{\%}, validating its general applicability. Finally, an ablation study further confirmed the significant impact of meta-data utilization on performance enhancement. We anticipate that our model can be universally applied to detect and filter low-quality documents in the real world.", }

Consistency within a document is a crucial feature indicative of its quality. Recently, within the vast amount of information produced across various media, there exists a significant number of low-quality documents that either lack internal consistency or contain content utterly unrelated to their headlines. Such low-quality documents induce fatigue in readers and undermine the credibility of the media source that provided them. Consequently, research to automatically detect these low-quality documents based on natural language processing is imperative. In this study, we introduce a hierarchical graph convolutional network (HGCN) that can detect internal inconsistencies within a document and incongruences between the title and body. Moreover, we constructed the Inconsistency Dataset, leveraging published news data and its meta-data, to train our model to detect document inconsistencies. Experimental results demonstrated that the HGCN achieved superior performance with an accuracy of 91.20{\%} on our constructed Inconsistency Dataset, outperforming other comparative models. Additionally, on the publicly available incongruent-related dataset, the proposed methodology demonstrated a performance of 92.00{\%}, validating its general applicability. Finally, an ablation study further confirmed the significant impact of meta-data utilization on performance enhancement. We anticipate that our model can be universally applied to detect and filter low-quality documents in the real world.

Hierarchical Graph Convolutional Network Approach for Detecting Low-Quality Documents

lrec-main.710

Poster

https://aclanthology.org/2024.lrec-main.711.bib

https://aclanthology.org/2024.lrec-main.711/

@inproceedings{man-etal-2024-hierarchical, title = "Hierarchical Selection of Important Context for Generative Event Causality Identification with Optimal Transports", author = "Man, Hieu and Nguyen, Chien Van and Ngo, Nghia Trung and Ngo, Linh and Dernoncourt, Franck and Nguyen, Thien Huu", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.711", pages = "8122--8132", abstract = "We study the problem of Event Causality Identification (ECI) that seeks to predict causal relation between event mentions in the text. In contrast to previous classification-based models, a few recent ECI methods have explored generative models to deliver state-of-the-art performance. However, such generative models cannot handle document-level ECI where long context between event mentions must be encoded to secure correct predictions. In addition, previous generative ECI methods tend to rely on external toolkits or human annotation to obtain necessary training signals. To address these limitations, we propose a novel generative framework that leverages Optimal Transport (OT) to automatically select the most important sentences and words from full documents. Specifically, we introduce hierarchical OT alignments between event pairs and the document to extract pertinent contexts. The selected sentences and words are provided as input and output to a T5 encoder-decoder model which is trained to generate both the causal relation label and salient contexts. This allows richer supervision without external tools. We conduct extensive evaluations on different datasets with multiple languages to demonstrate the benefits and state-of-the-art performance of ECI.", }

We study the problem of Event Causality Identification (ECI) that seeks to predict causal relation between event mentions in the text. In contrast to previous classification-based models, a few recent ECI methods have explored generative models to deliver state-of-the-art performance. However, such generative models cannot handle document-level ECI where long context between event mentions must be encoded to secure correct predictions. In addition, previous generative ECI methods tend to rely on external toolkits or human annotation to obtain necessary training signals. To address these limitations, we propose a novel generative framework that leverages Optimal Transport (OT) to automatically select the most important sentences and words from full documents. Specifically, we introduce hierarchical OT alignments between event pairs and the document to extract pertinent contexts. The selected sentences and words are provided as input and output to a T5 encoder-decoder model which is trained to generate both the causal relation label and salient contexts. This allows richer supervision without external tools. We conduct extensive evaluations on different datasets with multiple languages to demonstrate the benefits and state-of-the-art performance of ECI.

Hierarchical Selection of Important Context for Generative Event Causality Identification with Optimal Transports

lrec-main.711

Poster

https://aclanthology.org/2024.lrec-main.712.bib

https://aclanthology.org/2024.lrec-main.712/

@inproceedings{lin-etal-2024-hierarchical, title = "Hierarchical Topic Modeling via Contrastive Learning and Hyperbolic Embedding", author = "Lin, Zhicheng and Chen, HeGang and Lu, Yuyin and Rao, Yanghui and Xu, Hao and Lai, Hanjiang", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.712", pages = "8133--8143", abstract = "Hierarchical topic modeling, which can mine implicit semantics in the corpus and automatically construct topic hierarchical relationships, has received considerable attention recently. However, the current hierarchical topic models are mainly based on Euclidean space, which cannot well retain the implicit hierarchical semantic information in the corpus, leading to irrational structure of the generated topics. On the other hand, the existing Generative Adversarial Network (GAN) based neural topic models perform satisfactorily, but they remain constrained by pattern collapse due to the discontinuity of latent space. To solve the above problems, with the hypothesis of hyperbolic space, we propose a novel GAN-based hierarchical topic model to mine high-quality topics by introducing contrastive learning to capture information from documents. Furthermore, the distinct tree-like property of hyperbolic space preserves the implicit hierarchical semantics of documents in topic embeddings, which are projected into the hyperbolic space. Finally, we use a multi-head self-attention mechanism to learn implicit hierarchical semantics of topics and mine topic structure information. Experiments on real-world corpora demonstrate the remarkable performance of our model on topic coherence and topic diversity, as well as the rationality of the topic hierarchy.", }

Hierarchical topic modeling, which can mine implicit semantics in the corpus and automatically construct topic hierarchical relationships, has received considerable attention recently. However, the current hierarchical topic models are mainly based on Euclidean space, which cannot well retain the implicit hierarchical semantic information in the corpus, leading to irrational structure of the generated topics. On the other hand, the existing Generative Adversarial Network (GAN) based neural topic models perform satisfactorily, but they remain constrained by pattern collapse due to the discontinuity of latent space. To solve the above problems, with the hypothesis of hyperbolic space, we propose a novel GAN-based hierarchical topic model to mine high-quality topics by introducing contrastive learning to capture information from documents. Furthermore, the distinct tree-like property of hyperbolic space preserves the implicit hierarchical semantics of documents in topic embeddings, which are projected into the hyperbolic space. Finally, we use a multi-head self-attention mechanism to learn implicit hierarchical semantics of topics and mine topic structure information. Experiments on real-world corpora demonstrate the remarkable performance of our model on topic coherence and topic diversity, as well as the rationality of the topic hierarchy.

Hierarchical Topic Modeling via Contrastive Learning and Hyperbolic Embedding

lrec-main.712

Poster

https://aclanthology.org/2024.lrec-main.713.bib

https://aclanthology.org/2024.lrec-main.713/

@inproceedings{gu-etal-2024-high, title = "High-order Joint Constituency and Dependency Parsing", author = "Gu, Yanggan and Hou, Yang and Wang, Zhefeng and Duan, Xinyu and Li, Zhenghua", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.713", pages = "8144--8154", abstract = "This work revisits the topic of jointly parsing constituency and dependency trees, i.e., to produce compatible constituency and dependency trees simultaneously for input sentences, which is attractive considering that the two types of trees are complementary in representing syntax. The original work of Zhou and Zhao (2019) performs joint parsing only at the inference phase. They train two separate parsers under the multi-task learning framework (i.e., one shared encoder and two independent decoders). They design an ad-hoc dynamic programming-based decoding algorithm of $O(n^5)$ time complexity for finding optimal compatible tree pairs. Compared to their work, we make progress in three aspects: (1) adopting a much more efficient decoding algorithm of $O(n^4)$ time complexity, (2) exploring joint modeling at the training phase, instead of only at the inference phase, (3) proposing high-order scoring components to promote constituent-dependency interaction. We conduct experiments and analysis on seven languages, covering both rich-resource and low-resource scenarios. Results and analysis show that joint modeling leads to a modest overall performance boost over separate modeling, but substantially improves the complete matching ratio of whole trees, thanks to the explicit modeling of tree compatibility.", }

This work revisits the topic of jointly parsing constituency and dependency trees, i.e., to produce compatible constituency and dependency trees simultaneously for input sentences, which is attractive considering that the two types of trees are complementary in representing syntax. The original work of Zhou and Zhao (2019) performs joint parsing only at the inference phase. They train two separate parsers under the multi-task learning framework (i.e., one shared encoder and two independent decoders). They design an ad-hoc dynamic programming-based decoding algorithm of $O(n^5)$ time complexity for finding optimal compatible tree pairs. Compared to their work, we make progress in three aspects: (1) adopting a much more efficient decoding algorithm of $O(n^4)$ time complexity, (2) exploring joint modeling at the training phase, instead of only at the inference phase, (3) proposing high-order scoring components to promote constituent-dependency interaction. We conduct experiments and analysis on seven languages, covering both rich-resource and low-resource scenarios. Results and analysis show that joint modeling leads to a modest overall performance boost over separate modeling, but substantially improves the complete matching ratio of whole trees, thanks to the explicit modeling of tree compatibility.

High-order Joint Constituency and Dependency Parsing

lrec-main.713

Poster

2309.11888

https://aclanthology.org/2024.lrec-main.714.bib

https://aclanthology.org/2024.lrec-main.714/

@inproceedings{gao-etal-2024-high, title = "High-Order Semantic Alignment for Unsupervised Fine-Grained Image-Text Retrieval", author = "Gao, Rui and Cheng, Miaomiao and Han, Xu and Song, Wei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.714", pages = "8155--8165", abstract = "Cross-modal retrieval is an important yet challenging task due to the semantic discrepancy between visual content and language. To measure the correlation between images and text, most existing research mainly focuses on learning global or local correspondence, failing to explore fine-grained local-global alignment. To infer more accurate similarity scores, we introduce a novel High Order Semantic Alignment (HOSA) model that can provide complementary and comprehensive semantic clues. Specifically, to jointly learn global and local alignment and emphasize local-global interaction, we employ tensor-product (t-product) operation to reconstruct one modal{'}s representation based on another modal{'}s information in a common semantic space. Such a cross-modal reconstruction strategy would significantly enhance inter-modal correlation learning in a fine-grained manner. Extensive experiments on two benchmark datasets validate that our model significantly outperforms several state-of-the-art baselines, especially in retrieving the most relevant results.", }

Cross-modal retrieval is an important yet challenging task due to the semantic discrepancy between visual content and language. To measure the correlation between images and text, most existing research mainly focuses on learning global or local correspondence, failing to explore fine-grained local-global alignment. To infer more accurate similarity scores, we introduce a novel High Order Semantic Alignment (HOSA) model that can provide complementary and comprehensive semantic clues. Specifically, to jointly learn global and local alignment and emphasize local-global interaction, we employ tensor-product (t-product) operation to reconstruct one modal{'}s representation based on another modal{'}s information in a common semantic space. Such a cross-modal reconstruction strategy would significantly enhance inter-modal correlation learning in a fine-grained manner. Extensive experiments on two benchmark datasets validate that our model significantly outperforms several state-of-the-art baselines, especially in retrieving the most relevant results.

High-Order Semantic Alignment for Unsupervised Fine-Grained Image-Text Retrieval

lrec-main.714

Poster

https://aclanthology.org/2024.lrec-main.715.bib

https://aclanthology.org/2024.lrec-main.715/

@inproceedings{pavlopoulos-etal-2024-holm, title = "{H}o{LM}: Analyzing the Linguistic Unexpectedness in {H}omeric Poetry", author = "Pavlopoulos, John and Sandell, Ryan and Konstantinidou, Maria and Bozzone, Chiara", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.715", pages = "8166--8172", abstract = "The authorship of the Homeric poems has been a matter of debate for centuries. Computational approaches such as language modeling exist that can aid experts in making crucial headway. We observe, however, that such work has, thus far, only been carried out at the level of lengthier excerpts, but not individual verses, the level at which most suspected interpolations occur. We address this weakness by presenting a corpus of Homeric verses, each complemented with a score quantifying linguistic unexpectedness based on Perplexity. We assess the nature of these scores by exploring their correlation with named entities, the frequency of character n-grams, and (inverse) word frequency, revealing robust correlations with the latter two. This apparent bias can be partly overcome by simply dividing scores for unexpectedness by the maximum term frequency per verse.", }

The authorship of the Homeric poems has been a matter of debate for centuries. Computational approaches such as language modeling exist that can aid experts in making crucial headway. We observe, however, that such work has, thus far, only been carried out at the level of lengthier excerpts, but not individual verses, the level at which most suspected interpolations occur. We address this weakness by presenting a corpus of Homeric verses, each complemented with a score quantifying linguistic unexpectedness based on Perplexity. We assess the nature of these scores by exploring their correlation with named entities, the frequency of character n-grams, and (inverse) word frequency, revealing robust correlations with the latter two. This apparent bias can be partly overcome by simply dividing scores for unexpectedness by the maximum term frequency per verse.

HoLM: Analyzing the Linguistic Unexpectedness in Homeric Poetry

lrec-main.715

Poster

https://aclanthology.org/2024.lrec-main.716.bib

https://aclanthology.org/2024.lrec-main.716/

@inproceedings{zaczynska-etal-2024-diplomats, title = "How Diplomats Dispute: The {UN} Security Council Conflict Corpus", author = "Zaczynska, Karolina and Bourgonje, Peter and Stede, Manfred", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.716", pages = "8173--8183", abstract = "We investigate disputes in the United Nations Security Council (UNSC) by studying the linguistic means of expressing conflicts. As a result, we present the UNSC Conflict Corpus (UNSCon), a collection of 87 UNSC speeches that are annotated for conflicts. We explain and motivate our annotation scheme and report on a series of experiments for automatic conflict classification. Further, we demonstrate the difficulty when dealing with diplomatic language - which is highly complex and often implicit along various dimensions - by providing corpus examples, readability scores, and classification results.", }

We investigate disputes in the United Nations Security Council (UNSC) by studying the linguistic means of expressing conflicts. As a result, we present the UNSC Conflict Corpus (UNSCon), a collection of 87 UNSC speeches that are annotated for conflicts. We explain and motivate our annotation scheme and report on a series of experiments for automatic conflict classification. Further, we demonstrate the difficulty when dealing with diplomatic language - which is highly complex and often implicit along various dimensions - by providing corpus examples, readability scores, and classification results.

How Diplomats Dispute: The UN Security Council Conflict Corpus

lrec-main.716

Poster

https://aclanthology.org/2024.lrec-main.717.bib

https://aclanthology.org/2024.lrec-main.717/

@inproceedings{gaido-etal-2024-hyenas, title = "How Do Hyenas Deal with Human Speech? Speech Recognition and Translation with {C}onf{H}yena", author = "Gaido, Marco and Papi, Sara and Negri, Matteo and Bentivogli, Luisa", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.717", pages = "8184--8191", abstract = "The attention mechanism, a cornerstone of state-of-the-art neural models, faces computational hurdles in processing long sequences due to its quadratic complexity. Consequently, research efforts in the last few years focused on finding more efficient alternatives. Among them, Hyena (Poli et al., 2023) stands out for achieving competitive results in both language modeling and image classification, while offering sub-quadratic memory and computational complexity. Building on these promising results, we propose ConfHyena, a Conformer whose encoder self-attentions are replaced with an adaptation of Hyena for speech processing, where the long input sequences cause high computational costs. Through experiments in automatic speech recognition (for English) and translation (from English into 8 target languages), we show that our best ConfHyena model significantly reduces the training time by 27{\%}, at the cost of minimal quality degradation (∼1{\%}), which, in most cases, is not statistically significant.", }

The attention mechanism, a cornerstone of state-of-the-art neural models, faces computational hurdles in processing long sequences due to its quadratic complexity. Consequently, research efforts in the last few years focused on finding more efficient alternatives. Among them, Hyena (Poli et al., 2023) stands out for achieving competitive results in both language modeling and image classification, while offering sub-quadratic memory and computational complexity. Building on these promising results, we propose ConfHyena, a Conformer whose encoder self-attentions are replaced with an adaptation of Hyena for speech processing, where the long input sequences cause high computational costs. Through experiments in automatic speech recognition (for English) and translation (from English into 8 target languages), we show that our best ConfHyena model significantly reduces the training time by 27{\%}, at the cost of minimal quality degradation (∼1{\%}), which, in most cases, is not statistically significant.

How Do Hyenas Deal with Human Speech? Speech Recognition and Translation with ConfHyena

lrec-main.717

Poster

2402.13208

https://huggingface.co/papers/2402.13208

https://aclanthology.org/2024.lrec-main.718.bib

https://aclanthology.org/2024.lrec-main.718/

@inproceedings{kayal-etal-2024-far, title = "How Far Is Too Far? Studying the Effects of Domain Discrepancy on Masked Language Models", author = "Kayal, Subhradeep and Rakhlin, Alexander and Dashti, Ali and Stepaniants, Serguei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.718", pages = "8192--8199", abstract = "Pre-trained masked language models, such as BERT, perform strongly on a wide variety of NLP tasks and have become ubiquitous in recent years. The typical way to use such models is to fine-tune them on downstream data. In this work, we aim to study how the difference in domains between the pre-trained model and the task effects its final performance. We first devise a simple mechanism to quantify the domain difference (using a cloze task) and use it to partition our dataset. Using these partitions of varying domain discrepancy, we focus on answering key questions around the impact of discrepancy on final performance, robustness to out-of-domain test-time examples and effect of domain-adaptive pre-training. We base our experiments on a large-scale openly available e-commerce dataset, and our findings suggest that in spite of pre-training the performance of BERT degrades on datasets with high domain discrepancy, especially in low resource cases. This effect is somewhat mitigated by continued pre-training for domain adaptation. Furthermore, the domain-gap also makes BERT sensitive to out-of-domain examples during inference, even in high resource tasks, and it is prudent to use as diverse a dataset as possible during fine-tuning to make it robust to domain shift.", }

Pre-trained masked language models, such as BERT, perform strongly on a wide variety of NLP tasks and have become ubiquitous in recent years. The typical way to use such models is to fine-tune them on downstream data. In this work, we aim to study how the difference in domains between the pre-trained model and the task effects its final performance. We first devise a simple mechanism to quantify the domain difference (using a cloze task) and use it to partition our dataset. Using these partitions of varying domain discrepancy, we focus on answering key questions around the impact of discrepancy on final performance, robustness to out-of-domain test-time examples and effect of domain-adaptive pre-training. We base our experiments on a large-scale openly available e-commerce dataset, and our findings suggest that in spite of pre-training the performance of BERT degrades on datasets with high domain discrepancy, especially in low resource cases. This effect is somewhat mitigated by continued pre-training for domain adaptation. Furthermore, the domain-gap also makes BERT sensitive to out-of-domain examples during inference, even in high resource tasks, and it is prudent to use as diverse a dataset as possible during fine-tuning to make it robust to domain shift.

How Far Is Too Far? Studying the Effects of Domain Discrepancy on Masked Language Models

lrec-main.718

Poster

https://aclanthology.org/2024.lrec-main.719.bib

https://aclanthology.org/2024.lrec-main.719/

@inproceedings{al-ali-libovicky-2024-gender, title = "How Gender Interacts with Political Values: A Case Study on {C}zech {BERT} Models", author = "Al Ali, Adnan and Libovick{\'y}, Jind{\v{r}}ich", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.719", pages = "8200--8210", abstract = "Neural language models, which reach state-of-the-art results on most natural language processing tasks, are trained on large text corpora that inevitably contain value-burdened content and often capture undesirable biases, which the models reflect. This case study focuses on the political biases of pre-trained encoders in Czech and compares them with a representative value survey. Because Czech is a gendered language, we also measure how the grammatical gender coincides with responses to men and women in the survey. We introduce a novel method for measuring the model{'}s perceived political values. We find that the models do not assign statement probability following value-driven reasoning, and there is no systematic difference between feminine and masculine sentences. We conclude that BERT-sized models do not manifest systematic alignment with political values and that the biases observed in the models are rather due to superficial imitation of training data patterns than systematic value beliefs encoded in the models.", }

Neural language models, which reach state-of-the-art results on most natural language processing tasks, are trained on large text corpora that inevitably contain value-burdened content and often capture undesirable biases, which the models reflect. This case study focuses on the political biases of pre-trained encoders in Czech and compares them with a representative value survey. Because Czech is a gendered language, we also measure how the grammatical gender coincides with responses to men and women in the survey. We introduce a novel method for measuring the model{'}s perceived political values. We find that the models do not assign statement probability following value-driven reasoning, and there is no systematic difference between feminine and masculine sentences. We conclude that BERT-sized models do not manifest systematic alignment with political values and that the biases observed in the models are rather due to superficial imitation of training data patterns than systematic value beliefs encoded in the models.

How Gender Interacts with Political Values: A Case Study on Czech BERT Models

lrec-main.719

Poster

2403.13514

https://aclanthology.org/2024.lrec-main.720.bib

https://aclanthology.org/2024.lrec-main.720/

@inproceedings{liu-etal-2024-good, title = "How Good Are {LLM}s at Out-of-Distribution Detection?", author = "Liu, Bo and Zhan, Li-Ming and Lu, Zexin and Feng, Yujie and Xue, Lei and Wu, Xiao-Ming", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.720", pages = "8211--8222", abstract = "Out-of-distribution (OOD) detection plays a vital role in enhancing the reliability of machine learning models. As large language models (LLMs) become more prevalent, the applicability of prior research on OOD detection that utilized smaller-scale Transformers such as BERT, RoBERTa, and GPT-2 may be challenged, due to the significant differences in the scale of these models, their pre-training objectives, and the paradigms used for inference. This paper initiates a pioneering empirical investigation into the OOD detection capabilities of LLMs, focusing on the LLaMA series ranging from 7B to 65B in size. We thoroughly evaluate commonly used OOD detectors, examining their performance in both zero-grad and fine-tuning scenarios. Notably, we alter previous discriminative in-distribution fine-tuning into generative fine-tuning, aligning the pre-training objective of LLMs with downstream tasks. Our findings unveil that a simple cosine distance OOD detector demonstrates superior efficacy, outperforming other OOD detectors. We provide an intriguing explanation for this phenomenon by highlighting the isotropic nature of the embedding spaces of LLMs, which distinctly contrasts with the anisotropic property observed in smaller BERT family models. The new insight enhances our understanding of how LLMs detect OOD data, thereby enhancing their adaptability and reliability in dynamic environments. We have released the source code at \url{https://github.com/Awenbocc/LLM-OOD} for other researchers to reproduce our results.", }

Out-of-distribution (OOD) detection plays a vital role in enhancing the reliability of machine learning models. As large language models (LLMs) become more prevalent, the applicability of prior research on OOD detection that utilized smaller-scale Transformers such as BERT, RoBERTa, and GPT-2 may be challenged, due to the significant differences in the scale of these models, their pre-training objectives, and the paradigms used for inference. This paper initiates a pioneering empirical investigation into the OOD detection capabilities of LLMs, focusing on the LLaMA series ranging from 7B to 65B in size. We thoroughly evaluate commonly used OOD detectors, examining their performance in both zero-grad and fine-tuning scenarios. Notably, we alter previous discriminative in-distribution fine-tuning into generative fine-tuning, aligning the pre-training objective of LLMs with downstream tasks. Our findings unveil that a simple cosine distance OOD detector demonstrates superior efficacy, outperforming other OOD detectors. We provide an intriguing explanation for this phenomenon by highlighting the isotropic nature of the embedding spaces of LLMs, which distinctly contrasts with the anisotropic property observed in smaller BERT family models. The new insight enhances our understanding of how LLMs detect OOD data, thereby enhancing their adaptability and reliability in dynamic environments. We have released the source code at \url{https://github.com/Awenbocc/LLM-OOD} for other researchers to reproduce our results.

How Good Are LLMs at Out-of-Distribution Detection?

lrec-main.720

Poster

2308.10261

https://aclanthology.org/2024.lrec-main.721.bib

https://aclanthology.org/2024.lrec-main.721/

@inproceedings{labrak-etal-2024-important, title = "How Important Is Tokenization in {F}rench Medical Masked Language Models?", author = "Labrak, Yanis and Bazoge, Adrien and Daille, B{\'e}atrice and Rouvier, Mickael and Dufour, Richard", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.721", pages = "8223--8234", abstract = "Subword tokenization has become the prevailing standard in the field of natural language processing (NLP) over recent years, primarily due to the widespread utilization of pre-trained language models. This shift began with Byte-Pair Encoding (BPE) and was later followed by the adoption of SentencePiece and WordPiece. While subword tokenization consistently outperforms character and word-level tokenization, the precise factors contributing to its success remain unclear. Key aspects such as the optimal segmentation granularity for diverse tasks and languages, the influence of data sources on tokenizers, and the role of morphological information in Indo-European languages remain insufficiently explored. This is particularly pertinent for biomedical terminology, characterized by specific rules governing morpheme combinations. Despite the agglutinative nature of biomedical terminology, existing language models do not explicitly incorporate this knowledge, leading to inconsistent tokenization strategies for common terms. In this paper, we seek to delve into the complexities of subword tokenization in French biomedical domain across a variety of NLP tasks and pinpoint areas where further enhancements can be made. We analyze classical tokenization algorithms, including BPE and SentencePiece, and introduce an original tokenization strategy that integrates morpheme-enriched word segmentation into existing tokenization methods.", }

Subword tokenization has become the prevailing standard in the field of natural language processing (NLP) over recent years, primarily due to the widespread utilization of pre-trained language models. This shift began with Byte-Pair Encoding (BPE) and was later followed by the adoption of SentencePiece and WordPiece. While subword tokenization consistently outperforms character and word-level tokenization, the precise factors contributing to its success remain unclear. Key aspects such as the optimal segmentation granularity for diverse tasks and languages, the influence of data sources on tokenizers, and the role of morphological information in Indo-European languages remain insufficiently explored. This is particularly pertinent for biomedical terminology, characterized by specific rules governing morpheme combinations. Despite the agglutinative nature of biomedical terminology, existing language models do not explicitly incorporate this knowledge, leading to inconsistent tokenization strategies for common terms. In this paper, we seek to delve into the complexities of subword tokenization in French biomedical domain across a variety of NLP tasks and pinpoint areas where further enhancements can be made. We analyze classical tokenization algorithms, including BPE and SentencePiece, and introduce an original tokenization strategy that integrates morpheme-enriched word segmentation into existing tokenization methods.

How Important Is Tokenization in French Medical Masked Language Models?

lrec-main.721

Poster

2402.15010

https://aclanthology.org/2024.lrec-main.722.bib

https://aclanthology.org/2024.lrec-main.722/

@inproceedings{ju-etal-2024-large, title = "How Large Language Models Encode Context Knowledge? A Layer-Wise Probing Study", author = "Ju, Tianjie and Sun, Weiwei and Du, Wei and Yuan, Xinwei and Ren, Zhaochun and Liu, Gongshen", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.722", pages = "8235--8246", abstract = "Previous work has showcased the intriguing capability of large language models (LLMs) in retrieving facts and processing context knowledge. However, only limited research exists on the layer-wise capability of LLMs to encode knowledge, which challenges our understanding of their internal mechanisms. In this paper, we devote the first attempt to investigate the layer-wise capability of LLMs through probing tasks. We leverage the powerful generative capability of ChatGPT to construct probing datasets, providing diverse and coherent evidence corresponding to various facts. We employ $\mathcal V$-usable information as the validation metric to better reflect the capability in encoding context knowledge across different layers. Our experiments on conflicting and newly acquired knowledge show that LLMs: (1) prefer to encode more context knowledge in the upper layers; (2) primarily encode context knowledge within knowledge-related entity tokens at lower layers while progressively expanding more knowledge within other tokens at upper layers; and (3) gradually forget the earlier context knowledge retained within the intermediate layers when provided with irrelevant evidence. Code is publicly available at https://github.com/Jometeorie/probing{\_}llama.", }

Previous work has showcased the intriguing capability of large language models (LLMs) in retrieving facts and processing context knowledge. However, only limited research exists on the layer-wise capability of LLMs to encode knowledge, which challenges our understanding of their internal mechanisms. In this paper, we devote the first attempt to investigate the layer-wise capability of LLMs through probing tasks. We leverage the powerful generative capability of ChatGPT to construct probing datasets, providing diverse and coherent evidence corresponding to various facts. We employ $\mathcal V$-usable information as the validation metric to better reflect the capability in encoding context knowledge across different layers. Our experiments on conflicting and newly acquired knowledge show that LLMs: (1) prefer to encode more context knowledge in the upper layers; (2) primarily encode context knowledge within knowledge-related entity tokens at lower layers while progressively expanding more knowledge within other tokens at upper layers; and (3) gradually forget the earlier context knowledge retained within the intermediate layers when provided with irrelevant evidence. Code is publicly available at https://github.com/Jometeorie/probing{\_}llama.

How Large Language Models Encode Context Knowledge? A Layer-Wise Probing Study

lrec-main.722

Poster

2402.16061

https://huggingface.co/papers/2402.16061

https://aclanthology.org/2024.lrec-main.723.bib

https://aclanthology.org/2024.lrec-main.723/

@inproceedings{orme-etal-2024-much, title = "How Much Do Robots Understand Rudeness? Challenges in Human-Robot Interaction", author = "Orme, Michael Andrew and Yu, Yanchao and Tan, Zhiyuan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.723", pages = "8247--8257", abstract = "This paper concerns the pressing need to understand and manage inappropriate language within the evolving human-robot interaction (HRI) landscape. As intelligent systems and robots transition from controlled laboratory settings to everyday households, the demand for polite and culturally sensitive conversational abilities becomes paramount, especially for younger individuals. This study explores data cleaning methods, focussing on rudeness and contextual similarity, to identify and mitigate inappropriate language in real-time interactions. State-of-the-art natural language models are also evaluated for their proficiency in discerning rudeness. This multifaceted investigation highlights the challenges of handling inappropriate language, including its tendency to hide within idiomatic expressions and its context-dependent nature. This study will further contribute to the future development of AI systems capable of engaging in intelligent conversations and upholding the values of courtesy and respect across diverse cultural and generational boundaries.", }

This paper concerns the pressing need to understand and manage inappropriate language within the evolving human-robot interaction (HRI) landscape. As intelligent systems and robots transition from controlled laboratory settings to everyday households, the demand for polite and culturally sensitive conversational abilities becomes paramount, especially for younger individuals. This study explores data cleaning methods, focussing on rudeness and contextual similarity, to identify and mitigate inappropriate language in real-time interactions. State-of-the-art natural language models are also evaluated for their proficiency in discerning rudeness. This multifaceted investigation highlights the challenges of handling inappropriate language, including its tendency to hide within idiomatic expressions and its context-dependent nature. This study will further contribute to the future development of AI systems capable of engaging in intelligent conversations and upholding the values of courtesy and respect across diverse cultural and generational boundaries.

How Much Do Robots Understand Rudeness? Challenges in Human-Robot Interaction

lrec-main.723

Poster

https://aclanthology.org/2024.lrec-main.724.bib

https://aclanthology.org/2024.lrec-main.724/

@inproceedings{ghosh-etal-2024-robust, title = "How Robust Are the {QA} Models for Hybrid Scientific Tabular Data? A Study Using Customized Dataset", author = "Ghosh, Akash and Bathini, Venkata Sahith and Ganguly, Niloy and Goyal, Pawan and Singh, Mayank", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.724", pages = "8258--8264", abstract = "Question-answering (QA) on hybrid scientific tabular and textual data deals with scientific information, and relies on complex numerical reasoning. In recent years, while tabular QA has seen rapid progress, understanding their robustness on scientific information is lacking due to absence of any benchmark dataset. To investigate the robustness of the existing state-of-the-art QA models on scientific hybrid tabular data, we propose a new dataset, {``}SciTabQA{''}, consisting of 822 question-answer pairs from scientific tables and their descriptions. With the help of this dataset, we assess the state-of-the-art Tabular QA models based on their ability (i) to use heterogeneous information requiring both structured data (table) and unstructured data (text) and (ii) to perform complex scientific reasoning tasks. In essence, we check the capability of the models to interpret scientific tables and text. Our experiments show that {``}SciTabQA{''} is an innovative dataset to study question-answering over scientific heterogeneous data. We benchmark three state-of-the-art Tabular QA models, and find that the best F1 score is only 0.462.", }

Question-answering (QA) on hybrid scientific tabular and textual data deals with scientific information, and relies on complex numerical reasoning. In recent years, while tabular QA has seen rapid progress, understanding their robustness on scientific information is lacking due to absence of any benchmark dataset. To investigate the robustness of the existing state-of-the-art QA models on scientific hybrid tabular data, we propose a new dataset, {``}SciTabQA{''}, consisting of 822 question-answer pairs from scientific tables and their descriptions. With the help of this dataset, we assess the state-of-the-art Tabular QA models based on their ability (i) to use heterogeneous information requiring both structured data (table) and unstructured data (text) and (ii) to perform complex scientific reasoning tasks. In essence, we check the capability of the models to interpret scientific tables and text. Our experiments show that {``}SciTabQA{''} is an innovative dataset to study question-answering over scientific heterogeneous data. We benchmark three state-of-the-art Tabular QA models, and find that the best F1 score is only 0.462.

How Robust Are the QA Models for Hybrid Scientific Tabular Data? A Study Using Customized Dataset

lrec-main.724

Poster

https://aclanthology.org/2024.lrec-main.725.bib

https://aclanthology.org/2024.lrec-main.725/

@inproceedings{liu-etal-2024-speculative, title = "How Speculative Can Speculative Decoding Be?", author = "Liu, Zhuorui and Zhang, Chen and Song, Dawei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.725", pages = "8265--8275", abstract = "Large language models (LLMs) have drawn great attention from the field of natural language processing and beyond, due to their impressive capability of autoregressive modeling, yet bringing an obvious problem, i.e., the largely increased latency. An emerging idea to alleviate this problem is speculative decoding, which first uses a draft model to draft tokens autoregressively and then makes the target model verify these tokens in parallel. The draft model is typically smaller than the target model, and it essentially trades generation quality for speed. Thereby, speculative decoding can be viewed as a speculative game for the target model in term of verification failures. That is, the lengthy draft tokens proposed by the small draft models could fail in the verification stage. Naturally, a critical question arises: how speculative can speculative decoding be, or in other words, how small can an adequate draft model be and how large can an appropriate number of draft tokens be? This work aims to investigate these questions and demonstrate how the scale of the draft model and the number of draft tokens would have an impact on the overall latency of the speculative decoding. We theoretically show that neither of above two factors will be infinitely speculative. Namely, there is a certain turning point for each of them. We then empirically show that the scale of the draft model could be 10-20$\times$ smaller than the target model and the optimal number of draft tokens should lie in 3-5.", }

Large language models (LLMs) have drawn great attention from the field of natural language processing and beyond, due to their impressive capability of autoregressive modeling, yet bringing an obvious problem, i.e., the largely increased latency. An emerging idea to alleviate this problem is speculative decoding, which first uses a draft model to draft tokens autoregressively and then makes the target model verify these tokens in parallel. The draft model is typically smaller than the target model, and it essentially trades generation quality for speed. Thereby, speculative decoding can be viewed as a speculative game for the target model in term of verification failures. That is, the lengthy draft tokens proposed by the small draft models could fail in the verification stage. Naturally, a critical question arises: how speculative can speculative decoding be, or in other words, how small can an adequate draft model be and how large can an appropriate number of draft tokens be? This work aims to investigate these questions and demonstrate how the scale of the draft model and the number of draft tokens would have an impact on the overall latency of the speculative decoding. We theoretically show that neither of above two factors will be infinitely speculative. Namely, there is a certain turning point for each of them. We then empirically show that the scale of the draft model could be 10-20$\times$ smaller than the target model and the optimal number of draft tokens should lie in 3-5.

How Speculative Can Speculative Decoding Be?

lrec-main.725

Poster

https://aclanthology.org/2024.lrec-main.726.bib

https://aclanthology.org/2024.lrec-main.726/

@inproceedings{payandeh-etal-2024-susceptible, title = "How Susceptible Are {LLM}s to Logical Fallacies?", author = "Payandeh, Amirreza and Pluth, Dan and Hosier, Jordan and Xiao, Xuesu and Gurbani, Vijay K.", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.726", pages = "8276--8286", abstract = "This paper investigates the rational thinking capability of Large Language Models (LLMs) in multi-round argumentative debates by exploring the impact of fallacious arguments on their logical reasoning performance. More specifically, we present Logic Competence Measurement Benchmark (LOGICOM), a diagnostic benchmark to assess the robustness of LLMs against logical fallacies. LOGICOM involves two agents: a persuader and a debater engaging in a multi-round debate on a controversial topic, where the persuader tries to convince the debater of the correctness of its claim. First, LOGICOM assesses the potential of LLMs to change their opinions through reasoning. Then, it evaluates the debater{'}s performance in logical reasoning by contrasting the scenario where the persuader employs logical fallacies against one where logical reasoning is used. We use this benchmark to evaluate the performance of GPT-3.5 and GPT-4 using a dataset containing controversial topics, claims, and reasons supporting them. Our findings indicate that both GPT-3.5 and GPT-4 can adjust their opinion through reasoning. However, when presented with logical fallacies, GPT-3.5 and GPT-4 are erroneously convinced 41{\%} and 69{\%} more often, respectively, compared to when logical reasoning is used. Finally, we introduce a new dataset containing over 5k pairs of logical vs. fallacious arguments.", }

This paper investigates the rational thinking capability of Large Language Models (LLMs) in multi-round argumentative debates by exploring the impact of fallacious arguments on their logical reasoning performance. More specifically, we present Logic Competence Measurement Benchmark (LOGICOM), a diagnostic benchmark to assess the robustness of LLMs against logical fallacies. LOGICOM involves two agents: a persuader and a debater engaging in a multi-round debate on a controversial topic, where the persuader tries to convince the debater of the correctness of its claim. First, LOGICOM assesses the potential of LLMs to change their opinions through reasoning. Then, it evaluates the debater{'}s performance in logical reasoning by contrasting the scenario where the persuader employs logical fallacies against one where logical reasoning is used. We use this benchmark to evaluate the performance of GPT-3.5 and GPT-4 using a dataset containing controversial topics, claims, and reasons supporting them. Our findings indicate that both GPT-3.5 and GPT-4 can adjust their opinion through reasoning. However, when presented with logical fallacies, GPT-3.5 and GPT-4 are erroneously convinced 41{\%} and 69{\%} more often, respectively, compared to when logical reasoning is used. Finally, we introduce a new dataset containing over 5k pairs of logical vs. fallacious arguments.

How Susceptible Are LLMs to Logical Fallacies?

lrec-main.726

Poster

2308.09853

https://huggingface.co/papers/2308.09853

https://aclanthology.org/2024.lrec-main.727.bib

https://aclanthology.org/2024.lrec-main.727/

@inproceedings{reinig-etal-2024-politics, title = "How to Do Politics with Words: Investigating Speech Acts in Parliamentary Debates", author = "Reinig, Ines and Rehbein, Ines and Ponzetto, Simone Paolo", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.727", pages = "8287--8300", abstract = "This paper presents a new perspective on framing through the lens of speech acts and investigates how politicians make use of different pragmatic speech act functions in political debates. To that end, we created a new resource of German parliamentary debates, annotated with fine-grained speech act types. Our hierarchical annotation scheme distinguishes between cooperation and conflict communication, further structured into six subtypes, such as informative, declarative or argumentative-critical speech acts, with 14 fine-grained classes at the lowest level. We present classification baselines on our new data and show that the fine-grained classes in our schema can be predicted with an avg. F1 of around 82.0{\%}. We then use our classifier to analyse the use of speech acts in a large corpus of parliamentary debates over a time span from 2003{--}2023.", }

This paper presents a new perspective on framing through the lens of speech acts and investigates how politicians make use of different pragmatic speech act functions in political debates. To that end, we created a new resource of German parliamentary debates, annotated with fine-grained speech act types. Our hierarchical annotation scheme distinguishes between cooperation and conflict communication, further structured into six subtypes, such as informative, declarative or argumentative-critical speech acts, with 14 fine-grained classes at the lowest level. We present classification baselines on our new data and show that the fine-grained classes in our schema can be predicted with an avg. F1 of around 82.0{\%}. We then use our classifier to analyse the use of speech acts in a large corpus of parliamentary debates over a time span from 2003{--}2023.

How to Do Politics with Words: Investigating Speech Acts in Parliamentary Debates

lrec-main.727

Poster

https://aclanthology.org/2024.lrec-main.728.bib

https://aclanthology.org/2024.lrec-main.728/

@inproceedings{bassignana-etal-2024-encode, title = "How to Encode Domain Information in Relation Classification", author = "Bassignana, Elisa and Gascou, Viggo Unmack and Laustsen, Frida N{\o}hr and Kristensen, Gustav and Petersen, Marie Haahr and van der Goot, Rob and Plank, Barbara", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.728", pages = "8301--8306", abstract = "Current language models require a lot of training data to obtain high performance. For Relation Classification (RC), many datasets are domain-specific, so combining datasets to obtain better performance is non-trivial. We explore a multi-domain training setup for RC, and attempt to improve performance by encoding domain information. Our proposed models improve {\textgreater} 2 Macro-F1 against the baseline setup, and our analysis reveals that not all the labels benefit the same: The classes which occupy a similar space across domains (i.e., their interpretation is close across them, for example {``}physical{''}) benefit the least, while domain-dependent relations (e.g., {``}part-of{''}) improve the most when encoding domain information.", }

Current language models require a lot of training data to obtain high performance. For Relation Classification (RC), many datasets are domain-specific, so combining datasets to obtain better performance is non-trivial. We explore a multi-domain training setup for RC, and attempt to improve performance by encoding domain information. Our proposed models improve {\textgreater} 2 Macro-F1 against the baseline setup, and our analysis reveals that not all the labels benefit the same: The classes which occupy a similar space across domains (i.e., their interpretation is close across them, for example {``}physical{''}) benefit the least, while domain-dependent relations (e.g., {``}part-of{''}) improve the most when encoding domain information.

How to Encode Domain Information in Relation Classification

lrec-main.728

Poster

2404.13760

https://aclanthology.org/2024.lrec-main.729.bib

https://aclanthology.org/2024.lrec-main.729/

@inproceedings{hangya-fraser-2024-solve, title = "How to Solve Few-Shot Abusive Content Detection Using the Data We Actually Have", author = "Hangya, Viktor and Fraser, Alexander", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.729", pages = "8307--8322", abstract = "Due to the broad range of social media platforms, the requirements of abusive language detection systems are varied and ever-changing. Already a large set of annotated corpora with different properties and label sets were created, such as hate or misogyny detection, but the form and targets of abusive speech are constantly evolving. Since, the annotation of new corpora is expensive, in this work we leverage datasets we already have, covering a wide range of tasks related to abusive language detection. Our goal is to build models cheaply for a new target label set and/or language, using only a few training examples of the target domain. We propose a two-step approach: first we train our model in a multitask fashion. We then carry out few-shot adaptation to the target requirements. Our experiments show that using already existing datasets and only a few-shots of the target task the performance of models improve both monolingually and across languages. Our analysis also shows that our models acquire a general understanding of abusive language, since they improve the prediction of labels which are present only in the target dataset and can benefit from knowledge about labels which are not directly used for the target task.", }

Due to the broad range of social media platforms, the requirements of abusive language detection systems are varied and ever-changing. Already a large set of annotated corpora with different properties and label sets were created, such as hate or misogyny detection, but the form and targets of abusive speech are constantly evolving. Since, the annotation of new corpora is expensive, in this work we leverage datasets we already have, covering a wide range of tasks related to abusive language detection. Our goal is to build models cheaply for a new target label set and/or language, using only a few training examples of the target domain. We propose a two-step approach: first we train our model in a multitask fashion. We then carry out few-shot adaptation to the target requirements. Our experiments show that using already existing datasets and only a few-shots of the target task the performance of models improve both monolingually and across languages. Our analysis also shows that our models acquire a general understanding of abusive language, since they improve the prediction of labels which are present only in the target dataset and can benefit from knowledge about labels which are not directly used for the target task.

How to Solve Few-Shot Abusive Content Detection Using the Data We Actually Have

lrec-main.729

Poster

2305.14081

https://aclanthology.org/2024.lrec-main.730.bib

https://aclanthology.org/2024.lrec-main.730/

@inproceedings{luo-etal-2024-understand, title = "How to Understand {``}Support{''}? An Implicit-enhanced Causal Inference Approach for Weakly-supervised Phrase Grounding", author = "Luo, Jiamin and Zhao, Jianing and Wang, Jingjing and Zhou, Guodong", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.730", pages = "8323--8333", abstract = "Weakly-supervised Phrase Grounding (WPG) is an emerging task of inferring the fine-grained phrase-region matching, while merely leveraging the coarse-grained sentence-image pairs for training. However, existing studies on WPG largely ignore the implicit phrase-region matching relations, which are crucial for evaluating the capability of models in understanding the deep multimodal semantics. To this end, this paper proposes an Implicit-Enhanced Causal Inference (IECI) approach to address the challenges of modeling the implicit relations and highlighting them beyond the explicit. Specifically, this approach leverages both the intervention and counterfactual techniques to tackle the above two challenges respectively. Furthermore, a high-quality implicit-enhanced dataset is annotated to evaluate IECI and detailed evaluations show the great advantages of IECI over the state-of-the-art baselines. Particularly, we observe an interesting finding that IECI outperforms the advanced multimodal LLMs by a large margin on this implicit-enhanced dataset, which may facilitate more research to evaluate the multimodal LLMs in this direction.", }

Weakly-supervised Phrase Grounding (WPG) is an emerging task of inferring the fine-grained phrase-region matching, while merely leveraging the coarse-grained sentence-image pairs for training. However, existing studies on WPG largely ignore the implicit phrase-region matching relations, which are crucial for evaluating the capability of models in understanding the deep multimodal semantics. To this end, this paper proposes an Implicit-Enhanced Causal Inference (IECI) approach to address the challenges of modeling the implicit relations and highlighting them beyond the explicit. Specifically, this approach leverages both the intervention and counterfactual techniques to tackle the above two challenges respectively. Furthermore, a high-quality implicit-enhanced dataset is annotated to evaluate IECI and detailed evaluations show the great advantages of IECI over the state-of-the-art baselines. Particularly, we observe an interesting finding that IECI outperforms the advanced multimodal LLMs by a large margin on this implicit-enhanced dataset, which may facilitate more research to evaluate the multimodal LLMs in this direction.

How to Understand “Support”? An Implicit-enhanced Causal Inference Approach for Weakly-supervised Phrase Grounding

lrec-main.730

Poster

https://aclanthology.org/2024.lrec-main.731.bib

https://aclanthology.org/2024.lrec-main.731/

@inproceedings{urbizu-etal-2024-well, title = "How Well Can {BERT} Learn the Grammar of an Agglutinative and Flexible-Order Language? The Case of {B}asque.", author = "Urbizu, Gorka and Zulaika, Muitze and Saralegi, Xabier and Corral, Ander", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.731", pages = "8334--8348", abstract = "This work investigates the acquisition of formal linguistic competence by neural language models, hypothesizing that languages with complex grammar, such as Basque, present substantial challenges during the pre-training phase. Basque is distinguished by its complex morphology and flexible word order, potentially complicating grammar extraction. In our analysis, we evaluated the grammatical knowledge of BERT models trained under various pre-training configurations, considering factors such as corpus size, model size, number of epochs, and the use of lemmatization. To assess this grammatical knowledge, we constructed the BL2MP (Basque L2 student-based Minimal Pairs) test set. This test set consists of minimal pairs, each containing both a grammatically correct and an incorrect sentence, sourced from essays authored by students at different proficiency levels in the Basque language. Additionally, our analysis explores the difficulties in learning various grammatical phenomena, the challenges posed by flexible word order, and the influence of the student{'}s proficiency level on the difficulty of correcting grammar errors.", }

This work investigates the acquisition of formal linguistic competence by neural language models, hypothesizing that languages with complex grammar, such as Basque, present substantial challenges during the pre-training phase. Basque is distinguished by its complex morphology and flexible word order, potentially complicating grammar extraction. In our analysis, we evaluated the grammatical knowledge of BERT models trained under various pre-training configurations, considering factors such as corpus size, model size, number of epochs, and the use of lemmatization. To assess this grammatical knowledge, we constructed the BL2MP (Basque L2 student-based Minimal Pairs) test set. This test set consists of minimal pairs, each containing both a grammatically correct and an incorrect sentence, sourced from essays authored by students at different proficiency levels in the Basque language. Additionally, our analysis explores the difficulties in learning various grammatical phenomena, the challenges posed by flexible word order, and the influence of the student{'}s proficiency level on the difficulty of correcting grammar errors.

How Well Can BERT Learn the Grammar of an Agglutinative and Flexible-Order Language? The Case of Basque.

lrec-main.731

Poster

https://aclanthology.org/2024.lrec-main.732.bib

https://aclanthology.org/2024.lrec-main.732/

@inproceedings{yang-etal-2024-hs, title = "{HS}-{GC}: Holistic Semantic Embedding and Global Contrast for Effective Text Clustering", author = "Yang, Chen and Cao, Bin and Fan, Jing", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.732", pages = "8349--8359", abstract = "In this paper, we introduce Holistic Semantic Embedding and Global Contrast (HS-GC), an end-to-end approach to learn the instance- and cluster-level representation. Specifically, for instance-level representation learning, we introduce a new loss function that exploits different layers of semantic information in a deep neural network to provide a more holistic semantic text representation. Contrastive learning is applied to these representations to improve the model{'}s ability to represent text instances. Additionally, for cluster-level representation learning we propose two strategies that utilize global update to construct cluster centers from a global view. The extensive experimental evaluation on five text datasets shows that our method outperforms the state-of-the-art model. Particularly on the SearchSnippets dataset, our method leads by 4.4{\%} in normalized mutual information against the latest comparison method. On the StackOverflow and TREC datasets, our method improves the clustering accuracy of 5.9{\%} and 3.2{\%}, respectively.", }

In this paper, we introduce Holistic Semantic Embedding and Global Contrast (HS-GC), an end-to-end approach to learn the instance- and cluster-level representation. Specifically, for instance-level representation learning, we introduce a new loss function that exploits different layers of semantic information in a deep neural network to provide a more holistic semantic text representation. Contrastive learning is applied to these representations to improve the model{'}s ability to represent text instances. Additionally, for cluster-level representation learning we propose two strategies that utilize global update to construct cluster centers from a global view. The extensive experimental evaluation on five text datasets shows that our method outperforms the state-of-the-art model. Particularly on the SearchSnippets dataset, our method leads by 4.4{\%} in normalized mutual information against the latest comparison method. On the StackOverflow and TREC datasets, our method improves the clustering accuracy of 5.9{\%} and 3.2{\%}, respectively.

HS-GC: Holistic Semantic Embedding and Global Contrast for Effective Text Clustering

lrec-main.732

Poster

https://aclanthology.org/2024.lrec-main.733.bib

https://aclanthology.org/2024.lrec-main.733/

@inproceedings{ligeti-nagy-etal-2024-hulu, title = "{H}u{LU}: {H}ungarian Language Understanding Benchmark Kit", author = "Ligeti-Nagy, No{\'e}mi and Ferenczi, Gerg{\H{o}} and H{\'e}ja, Enik{\H{o}} and Laki, L{\'a}szl{\'o} J{\'a}nos and Vad{\'a}sz, No{\'e}mi and Yang, Zijian Gy{\H{o}}z{\H{o}} and V{\'a}radi, Tam{\'a}s", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.733", pages = "8360--8371", abstract = "The paper introduces the Hungarian Language Understanding (HuLU) benchmark, a comprehensive assessment framework designed to evaluate the performance of neural language models on Hungarian language tasks. Inspired by the renowned GLUE and SuperGLUE benchmarks, HuLU aims to address the challenges specific to Hungarian language processing. The benchmark consists of various datasets, each representing different linguistic phenomena and task complexities. Moreover, the paper presents a web service developed for HuLU, offering a user-friendly interface for model evaluation. This platform not only ensures consistent assessment but also fosters transparency by maintaining a leaderboard showcasing model performances. Preliminary evaluations of various LMMs on HuLU datasets indicate that while Hungarian models show promise, there{'}s room for improvement to match the proficiency of English-centric models in their native language.", }

The paper introduces the Hungarian Language Understanding (HuLU) benchmark, a comprehensive assessment framework designed to evaluate the performance of neural language models on Hungarian language tasks. Inspired by the renowned GLUE and SuperGLUE benchmarks, HuLU aims to address the challenges specific to Hungarian language processing. The benchmark consists of various datasets, each representing different linguistic phenomena and task complexities. Moreover, the paper presents a web service developed for HuLU, offering a user-friendly interface for model evaluation. This platform not only ensures consistent assessment but also fosters transparency by maintaining a leaderboard showcasing model performances. Preliminary evaluations of various LMMs on HuLU datasets indicate that while Hungarian models show promise, there{'}s room for improvement to match the proficiency of English-centric models in their native language.

HuLU: Hungarian Language Understanding Benchmark Kit

lrec-main.733

Poster

https://aclanthology.org/2024.lrec-main.734.bib

https://aclanthology.org/2024.lrec-main.734/

@inproceedings{maladry-etal-2024-human, title = "Human and System Perspectives on the Expression of Irony: An Analysis of Likelihood Labels and Rationales", author = "Maladry, Aaron and Cignarella, Alessandra Teresa and Lefever, Els and van Hee, Cynthia and Hoste, Veronique", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.734", pages = "8372--8382", abstract = "In this paper, we examine the recognition of irony by both humans and automatic systems. We achieve this by enhancing the annotations of an English benchmark data set for irony detection. This enhancement involves a layer of human-annotated irony likelihood using a 7-point Likert scale that combines binary annotation with a confidence measure. Additionally, the annotators indicated the trigger words that led them to perceive the text as ironic, which leveraged necessary theoretical insights into the definition of irony and its various forms. By comparing these trigger word spans across annotators, we determine the extent to which humans agree on the source of irony in a text. Finally, we compare the human-annotated spans with sub-token importance attributions for fine-tuned transformers using Layer Integrated Gradients, a state-of-the-art interpretability metric. Our results indicate that our model achieves better performance on tweets that were annotated with high confidence and high agreement. Although automatic systems can identify trigger words with relative success, they still attribute a significant amount of their importance to the wrong tokens.", }

In this paper, we examine the recognition of irony by both humans and automatic systems. We achieve this by enhancing the annotations of an English benchmark data set for irony detection. This enhancement involves a layer of human-annotated irony likelihood using a 7-point Likert scale that combines binary annotation with a confidence measure. Additionally, the annotators indicated the trigger words that led them to perceive the text as ironic, which leveraged necessary theoretical insights into the definition of irony and its various forms. By comparing these trigger word spans across annotators, we determine the extent to which humans agree on the source of irony in a text. Finally, we compare the human-annotated spans with sub-token importance attributions for fine-tuned transformers using Layer Integrated Gradients, a state-of-the-art interpretability metric. Our results indicate that our model achieves better performance on tweets that were annotated with high confidence and high agreement. Although automatic systems can identify trigger words with relative success, they still attribute a significant amount of their importance to the wrong tokens.

Human and System Perspectives on the Expression of Irony: An Analysis of Likelihood Labels and Rationales

lrec-main.734

Poster

https://aclanthology.org/2024.lrec-main.735.bib

https://aclanthology.org/2024.lrec-main.735/

@inproceedings{peng-etal-2024-humaneval, title = "{H}uman{E}val-{XL}: A Multilingual Code Generation Benchmark for Cross-lingual Natural Language Generalization", author = "Peng, Qiwei and Chai, Yekun and Li, Xuhong", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.735", pages = "8383--8394", abstract = "Large language models (LLMs) have made significant progress in generating codes from textual prompts. However, existing benchmarks have mainly concentrated on translating English prompts to multilingual codes or have been constrained to very limited natural languages (NLs). These benchmarks have overlooked the vast landscape of massively multilingual NL to multilingual code, leaving a critical gap in the evaluation of multilingual LLMs. In response, we introduce HumanEval-XL, a massively multilingual code generation benchmark specifically crafted to address this deficiency. HumanEval-XL establishes connections between 23 NLs and 12 programming languages (PLs), and comprises of a collection of 22,080 prompts with an average of 8.33 test cases. By ensuring parallel data across multiple NLs and PLs, HumanEval-XL offers a comprehensive evaluation platform for multilingual LLMs, allowing the assessment of the understanding of different NLs. Our work serves as a pioneering step towards filling the void in evaluating NL generalization in the area of multilingual code generation. We make our evaluation code and data publicly available at https://github.com/FloatAI/HumanEval-XL.", }

Large language models (LLMs) have made significant progress in generating codes from textual prompts. However, existing benchmarks have mainly concentrated on translating English prompts to multilingual codes or have been constrained to very limited natural languages (NLs). These benchmarks have overlooked the vast landscape of massively multilingual NL to multilingual code, leaving a critical gap in the evaluation of multilingual LLMs. In response, we introduce HumanEval-XL, a massively multilingual code generation benchmark specifically crafted to address this deficiency. HumanEval-XL establishes connections between 23 NLs and 12 programming languages (PLs), and comprises of a collection of 22,080 prompts with an average of 8.33 test cases. By ensuring parallel data across multiple NLs and PLs, HumanEval-XL offers a comprehensive evaluation platform for multilingual LLMs, allowing the assessment of the understanding of different NLs. Our work serves as a pioneering step towards filling the void in evaluating NL generalization in the area of multilingual code generation. We make our evaluation code and data publicly available at https://github.com/FloatAI/HumanEval-XL.

HumanEval-XL: A Multilingual Code Generation Benchmark for Cross-lingual Natural Language Generalization

lrec-main.735

Poster

2402.16694

https://huggingface.co/papers/2402.16694

https://aclanthology.org/2024.lrec-main.736.bib

https://aclanthology.org/2024.lrec-main.736/

@inproceedings{thielmann-etal-2024-human, title = "Human in the Loop: How to Effectively Create Coherent Topics by Manually Labeling Only a Few Documents per Class", author = {Thielmann, Anton F. and Weisser, Christoph and S{\"a}fken, Benjamin}, editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.736", pages = "8395--8405", abstract = "Few-shot methods for accurate modeling under sparse label-settings have improved significantly. However, the applications of few-shot modeling in natural language processing remain solely in the field of document classification. With recent performance improvements, supervised few-shot methods, combined with a simple topic extraction method pose a significant challenge to unsupervised topic modeling methods. Our research shows that supervised few-shot learning, combined with a simple topic extraction method, can outperform unsupervised topic modeling techniques in terms of generating coherent topics, even when only a few labeled documents per class are used. The code is available at the following link: https://github.com/AnFreTh/STREAM", }

Few-shot methods for accurate modeling under sparse label-settings have improved significantly. However, the applications of few-shot modeling in natural language processing remain solely in the field of document classification. With recent performance improvements, supervised few-shot methods, combined with a simple topic extraction method pose a significant challenge to unsupervised topic modeling methods. Our research shows that supervised few-shot learning, combined with a simple topic extraction method, can outperform unsupervised topic modeling techniques in terms of generating coherent topics, even when only a few labeled documents per class are used. The code is available at the following link: https://github.com/AnFreTh/STREAM

Human in the Loop: How to Effectively Create Coherent Topics by Manually Labeling Only a Few Documents per Class

lrec-main.736

Poster

2212.09422

https://aclanthology.org/2024.lrec-main.737.bib

https://aclanthology.org/2024.lrec-main.737/

@inproceedings{wong-etal-2024-humanistic, title = "Humanistic Buddhism Corpus: A Challenging Domain-Specific Dataset of {E}nglish Translations for Classical and {M}odern {C}hinese", author = "Wong, Youheng W. and Parde, Natalie and Koyuncu, Erdem", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.737", pages = "8406--8417", abstract = "We introduce the Humanistic Buddhism Corpus (HBC), a dataset containing over 80,000 Chinese-English parallel phrases extracted and translated from publications in the domain of Buddhism. HBC is one of the largest free domain-specific datasets that is publicly available for research, containing text from both classical and modern Chinese. Moreover, since HBC originates from religious texts, many phrases in the dataset contain metaphors and symbolism, and are subject to multiple interpretations. Compared to existing machine translation datasets, HBC presents difficult unique challenges. In this paper, we describe HBC in detail. We evaluate HBC within a machine translation setting, validating its use by establishing performance benchmarks using a Transformer model with different transfer learning setups.", }

We introduce the Humanistic Buddhism Corpus (HBC), a dataset containing over 80,000 Chinese-English parallel phrases extracted and translated from publications in the domain of Buddhism. HBC is one of the largest free domain-specific datasets that is publicly available for research, containing text from both classical and modern Chinese. Moreover, since HBC originates from religious texts, many phrases in the dataset contain metaphors and symbolism, and are subject to multiple interpretations. Compared to existing machine translation datasets, HBC presents difficult unique challenges. In this paper, we describe HBC in detail. We evaluate HBC within a machine translation setting, validating its use by establishing performance benchmarks using a Transformer model with different transfer learning setups.

Humanistic Buddhism Corpus: A Challenging Domain-Specific Dataset of English Translations for Classical and Modern Chinese

lrec-main.737

Poster

https://aclanthology.org/2024.lrec-main.738.bib

https://aclanthology.org/2024.lrec-main.738/

@inproceedings{isaacs-etal-2024-humanitarian, title = "Humanitarian Corpora for {E}nglish, {F}rench and {S}panish", author = "Isaacs, Loryn and Chamb{\'o}, Santiago and Le{\'o}n-Ara{\'u}z, Pilar", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.738", pages = "8418--8426", abstract = "This paper presents three corpora of English, French and Spanish humanitarian documents compiled with reports obtained from ReliefWeb through its API. ReliefWeb is a leading database of humanitarian documents operated by the UN Office for the Coordination of Humanitarian Affairs (OCHA). To compile these corpora, documents were selected with language identification and noise reduction techniques. They were subsequently tokenized, lemmatized, tagged by part of speech, and enriched with metadata for use by linguists in corpus query software. These corpora were compiled to satisfy the research needs of the Humanitarian Encyclopedia, a project with a focus on conceptual variation. However, they can also be useful for other humanitarian endeavors, whether they are research- or practitioner-oriented; the source code for generating the corpora is available on GitHub. To compare materials, an exploratory analysis of definitional and generic-specific information was conducted for the concept of ARMED ACTOR with lexical data extracted from an English legacy corpus (where the concept is underrepresented) as well as on the new English and Spanish corpora. Lexical data were compared among corpora and presented by means of online data visualization to illustrate its potential to inform conceptual modelling.", }

This paper presents three corpora of English, French and Spanish humanitarian documents compiled with reports obtained from ReliefWeb through its API. ReliefWeb is a leading database of humanitarian documents operated by the UN Office for the Coordination of Humanitarian Affairs (OCHA). To compile these corpora, documents were selected with language identification and noise reduction techniques. They were subsequently tokenized, lemmatized, tagged by part of speech, and enriched with metadata for use by linguists in corpus query software. These corpora were compiled to satisfy the research needs of the Humanitarian Encyclopedia, a project with a focus on conceptual variation. However, they can also be useful for other humanitarian endeavors, whether they are research- or practitioner-oriented; the source code for generating the corpora is available on GitHub. To compare materials, an exploratory analysis of definitional and generic-specific information was conducted for the concept of ARMED ACTOR with lexical data extracted from an English legacy corpus (where the concept is underrepresented) as well as on the new English and Spanish corpora. Lexical data were compared among corpora and presented by means of online data visualization to illustrate its potential to inform conceptual modelling.

Humanitarian Corpora for English, French and Spanish

lrec-main.738

Poster

https://aclanthology.org/2024.lrec-main.739.bib

https://aclanthology.org/2024.lrec-main.739/

@inproceedings{zhou-etal-2024-humanizing, title = "Humanizing Machine-Generated Content: Evading {AI}-Text Detection through Adversarial Attack", author = "Zhou, Ying and He, Ben and Sun, Le", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.739", pages = "8427--8437", abstract = "With the development of large language models (LLMs), detecting whether text is generated by a machine becomes increasingly challenging in the face of malicious use cases like the spread of false information, protection of intellectual property, and prevention of academic plagiarism. While well-trained text detectors have demonstrated promising performance on unseen test data, recent research suggests that these detectors have vulnerabilities when dealing with adversarial attacks, such as paraphrasing. In this paper, we propose a framework for a broader class of adversarial attacks, designed to perform minor perturbations in machine-generated content to evade detection. We consider two attack settings: white-box and black-box, and employ adversarial learning in dynamic scenarios to assess the potential enhancement of the current detection model{'}s robustness against such attacks. The empirical results reveal that the current detection model can be compromised in as little as 10 seconds, leading to the misclassification of machine-generated text as human-written content. Furthermore, we explore the prospect of improving the model{'}s robustness over iterative adversarial learning. Although some improvements in model robustness are observed, practical applications still face significant challenges. These findings shed light on the future development of AI-text detectors, emphasizing the need for more accurate and robust detection methods.", }

With the development of large language models (LLMs), detecting whether text is generated by a machine becomes increasingly challenging in the face of malicious use cases like the spread of false information, protection of intellectual property, and prevention of academic plagiarism. While well-trained text detectors have demonstrated promising performance on unseen test data, recent research suggests that these detectors have vulnerabilities when dealing with adversarial attacks, such as paraphrasing. In this paper, we propose a framework for a broader class of adversarial attacks, designed to perform minor perturbations in machine-generated content to evade detection. We consider two attack settings: white-box and black-box, and employ adversarial learning in dynamic scenarios to assess the potential enhancement of the current detection model{'}s robustness against such attacks. The empirical results reveal that the current detection model can be compromised in as little as 10 seconds, leading to the misclassification of machine-generated text as human-written content. Furthermore, we explore the prospect of improving the model{'}s robustness over iterative adversarial learning. Although some improvements in model robustness are observed, practical applications still face significant challenges. These findings shed light on the future development of AI-text detectors, emphasizing the need for more accurate and robust detection methods.

Humanizing Machine-Generated Content: Evading AI-Text Detection through Adversarial Attack

lrec-main.739

Poster

2404.01907

https://aclanthology.org/2024.lrec-main.740.bib

https://aclanthology.org/2024.lrec-main.740/

@inproceedings{bourgeade-etal-2024-humans, title = "Humans Need Context, What about Machines? Investigating Conversational Context in Abusive Language Detection", author = "Bourgeade, Tom and Li, Zongmin and Benamara, Farah and Moriceau, V{\'e}ronique and Su, Jian and Sun, Aixin", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.740", pages = "8438--8452", abstract = "A crucial aspect in abusive language on social media platforms (toxicity, hate speech, harmful stereotypes, etc.) is its inherent contextual nature. In this paper, we focus on the role of conversational context in abusive language detection, one of the most {``}direct{''} forms of context in this domain, as given by the conversation threads (e.g., directly preceding message, original post). The incorporation of surrounding messages has proven vital for the accurate human annotation of harmful content. However, many prior works have either ignored this aspect, collecting and processing messages in isolation, or have obtained inconsistent results when attempting to embed such contextual information into traditional classification methods. The reasons behind these findings have not yet been properly addressed. To this end, we propose an analysis of the impact of conversational context in abusive language detection, through: (1) an analysis of prior works and the limitations of the most common concatenation-based approach, which we attempt to address with two alternative architectures; (2) an evaluation of these methods on existing datasets in English, and a new dataset of French tweets annotated for hate speech and stereotypes; and (3) a qualitative analysis showcasing the necessity for context-awareness in ALD, but also its difficulties.", }

A crucial aspect in abusive language on social media platforms (toxicity, hate speech, harmful stereotypes, etc.) is its inherent contextual nature. In this paper, we focus on the role of conversational context in abusive language detection, one of the most {``}direct{''} forms of context in this domain, as given by the conversation threads (e.g., directly preceding message, original post). The incorporation of surrounding messages has proven vital for the accurate human annotation of harmful content. However, many prior works have either ignored this aspect, collecting and processing messages in isolation, or have obtained inconsistent results when attempting to embed such contextual information into traditional classification methods. The reasons behind these findings have not yet been properly addressed. To this end, we propose an analysis of the impact of conversational context in abusive language detection, through: (1) an analysis of prior works and the limitations of the most common concatenation-based approach, which we attempt to address with two alternative architectures; (2) an evaluation of these methods on existing datasets in English, and a new dataset of French tweets annotated for hate speech and stereotypes; and (3) a qualitative analysis showcasing the necessity for context-awareness in ALD, but also its difficulties.

Humans Need Context, What about Machines? Investigating Conversational Context in Abusive Language Detection

lrec-main.740

Poster

https://aclanthology.org/2024.lrec-main.741.bib

https://aclanthology.org/2024.lrec-main.741/

@inproceedings{schmeisser-nieto-etal-2024-human, title = "Human vs. Machine Perceptions on Immigration Stereotypes", author = "Schmeisser-Nieto, Wolfgang S. and Pastells, Pol and Frenda, Simona and Taule, Mariona", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.741", pages = "8453--8463", abstract = "The increasing popularity of natural language processing has led to a race to improve machine learning models that often leaves aside the core study object, the language itself. In this study, we present classification models designed to detect stereotypes related to immigrants, along with both quantitative and qualitative analyses, shedding light on linguistic distinctions in how humans and various models perceive stereotypes. Given the subjective nature of this task, one of the models incorporates the judgments of all annotators by utilizing soft labels. Through a comparative analysis of BERT-based models using both hard and soft labels, along with predictions from GPT-4, we gain a clearer understanding of the linguistic challenges posed by texts containing stereotypes. Our dataset comprises Spanish Twitter posts collected as responses to immigrant-related hoaxes, annotated with binary values indicating the presence of stereotypes, implicitness, and the requirement for conversational context to understand the stereotype. Our findings suggest that both model prediction confidence and inter-annotator agreement are higher for explicit stereotypes, while stereotypes conveyed through irony and other figures of speech prove more challenging to detect than other implicit stereotypes.", }

The increasing popularity of natural language processing has led to a race to improve machine learning models that often leaves aside the core study object, the language itself. In this study, we present classification models designed to detect stereotypes related to immigrants, along with both quantitative and qualitative analyses, shedding light on linguistic distinctions in how humans and various models perceive stereotypes. Given the subjective nature of this task, one of the models incorporates the judgments of all annotators by utilizing soft labels. Through a comparative analysis of BERT-based models using both hard and soft labels, along with predictions from GPT-4, we gain a clearer understanding of the linguistic challenges posed by texts containing stereotypes. Our dataset comprises Spanish Twitter posts collected as responses to immigrant-related hoaxes, annotated with binary values indicating the presence of stereotypes, implicitness, and the requirement for conversational context to understand the stereotype. Our findings suggest that both model prediction confidence and inter-annotator agreement are higher for explicit stereotypes, while stereotypes conveyed through irony and other figures of speech prove more challenging to detect than other implicit stereotypes.

Human vs. Machine Perceptions on Immigration Stereotypes

lrec-main.741

Poster

https://aclanthology.org/2024.lrec-main.742.bib

https://aclanthology.org/2024.lrec-main.742/

@inproceedings{nuo-guo-2024-hybrid, title = "Hybrid of Spans and Table-Filling for Aspect-Level Sentiment Triplet Extraction", author = "Nuo, Minghua and Guo, Chaofan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.742", pages = "8464--8473", abstract = "Aspect Sentiment Triplet Extraction (ASTE) has become an emerging task in sentiment analysis research. Recently, researchers have proposed different tagging schemes, containing tagging of words, tagging of word pairs, and tagging of spans. However, the first two of these methods are often insufficient for the identification of multi-word terms, while the span tagging can label the entire phrase span, but it lacks the interactive information between words. In this paper, we propose Span in Table(S{\&}T) model which combining span with table-filling. Specifically, S{\&}T model achieve full fusion of syntactic and contextual features through cross-attention and generate the structures of word-pair table through Biaffine. Then, our model converts it to a span table by computing semantic distance based on syntactic dependency tree, which can enrich each unit of span table with semantic and interactive information. Meanwhile, the initial sentence features are constructed as simple phrase tables to enhance textual information of the phrase itself. In decoding, we define 8 types of labels for identifying three dimensions including aspect, opinion, and sentiment. Finally, the extensive experiments on D2 dataset show S{\&}T model achieves competitive results in ASTE task, the results certify the effectiveness and robustness of our S{\&}T model.", }

Aspect Sentiment Triplet Extraction (ASTE) has become an emerging task in sentiment analysis research. Recently, researchers have proposed different tagging schemes, containing tagging of words, tagging of word pairs, and tagging of spans. However, the first two of these methods are often insufficient for the identification of multi-word terms, while the span tagging can label the entire phrase span, but it lacks the interactive information between words. In this paper, we propose Span in Table(S{\&}T) model which combining span with table-filling. Specifically, S{\&}T model achieve full fusion of syntactic and contextual features through cross-attention and generate the structures of word-pair table through Biaffine. Then, our model converts it to a span table by computing semantic distance based on syntactic dependency tree, which can enrich each unit of span table with semantic and interactive information. Meanwhile, the initial sentence features are constructed as simple phrase tables to enhance textual information of the phrase itself. In decoding, we define 8 types of labels for identifying three dimensions including aspect, opinion, and sentiment. Finally, the extensive experiments on D2 dataset show S{\&}T model achieves competitive results in ASTE task, the results certify the effectiveness and robustness of our S{\&}T model.

Hybrid of Spans and Table-Filling for Aspect-Level Sentiment Triplet Extraction

lrec-main.742

Poster

https://aclanthology.org/2024.lrec-main.743.bib

https://aclanthology.org/2024.lrec-main.743/

@inproceedings{li-etal-2024-hyperbolic, title = "Hyperbolic Graph Neural Network for Temporal Knowledge Graph Completion", author = "Li, Yancong and Zhang, Xiaoming and Cui, Ying and Ma, Shuai", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.743", pages = "8474--8486", abstract = "Temporal Knowledge Graphs (TKGs) represent a crucial source of structured temporal information and exhibit significant utility in various real-world applications. However, TKGs are susceptible to incompleteness, necessitating Temporal Knowledge Graph Completion (TKGC) to predict missing facts. Existing models have encountered limitations in effectively capturing the intricate temporal dynamics and hierarchical relations within TKGs. To address these challenges, HyGNet is proposed, leveraging hyperbolic geometry to effectively model temporal knowledge graphs. The model comprises two components: the Hyperbolic Gated Graph Neural Network (HGGNN) and the Hyperbolic Convolutional Neural Network (HCNN). HGGNN aggregates neighborhood information in hyperbolic space, effectively capturing the contextual information and dependencies between entities. HCNN interacts with embeddings in hyperbolic space, effectively modeling the complex interactions between entities, relations, and timestamps. Additionally, a consistency loss is introduced to ensure smooth transitions in temporal embeddings. The extensive experimental results conducted on four benchmark datasets for TKGC highlight the effectiveness of HyGNet. It achieves state-of-the-art performance in comparison to previous models, showcasing its potential for real-world applications that involve temporal reasoning and knowledge prediction.", }

Temporal Knowledge Graphs (TKGs) represent a crucial source of structured temporal information and exhibit significant utility in various real-world applications. However, TKGs are susceptible to incompleteness, necessitating Temporal Knowledge Graph Completion (TKGC) to predict missing facts. Existing models have encountered limitations in effectively capturing the intricate temporal dynamics and hierarchical relations within TKGs. To address these challenges, HyGNet is proposed, leveraging hyperbolic geometry to effectively model temporal knowledge graphs. The model comprises two components: the Hyperbolic Gated Graph Neural Network (HGGNN) and the Hyperbolic Convolutional Neural Network (HCNN). HGGNN aggregates neighborhood information in hyperbolic space, effectively capturing the contextual information and dependencies between entities. HCNN interacts with embeddings in hyperbolic space, effectively modeling the complex interactions between entities, relations, and timestamps. Additionally, a consistency loss is introduced to ensure smooth transitions in temporal embeddings. The extensive experimental results conducted on four benchmark datasets for TKGC highlight the effectiveness of HyGNet. It achieves state-of-the-art performance in comparison to previous models, showcasing its potential for real-world applications that involve temporal reasoning and knowledge prediction.

Hyperbolic Graph Neural Network for Temporal Knowledge Graph Completion

lrec-main.743

Poster

https://aclanthology.org/2024.lrec-main.744.bib

https://aclanthology.org/2024.lrec-main.744/

@inproceedings{chen-etal-2024-hyperbolic, title = "Hyperbolic Representations for Prompt Learning", author = "Chen, Nan and Su, Xiangdong and Bao, Feilong", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.744", pages = "8487--8492", abstract = "Continuous prompt tuning has gained significant attention for its ability to train only continuous prompts while freezing the language model. This approach greatly reduces the training time and storage for downstream tasks. In this work, we delve into the hierarchical relationship between the prompts and downstream text inputs. In prompt learning, the prefix prompt acts as a module to guide the downstream language model, establishing a hierarchical relationship between the prefix prompt and subsequent inputs. Furthermore, we explore the benefits of leveraging hyperbolic space for modeling hierarchical structures. We project representations of pre-trained models from Euclidean space into hyperbolic space using the Poincar{\'e} disk which effectively captures the hierarchical relationship between the prompt and input text. The experiments on natural language understanding (NLU) tasks illustrate that hyperbolic space can model the hierarchical relationship between prompt and text input. We release our code at \url{https://github.com/myaxxxxx/Hyperbolic-Prompt-Learning}.", }

Continuous prompt tuning has gained significant attention for its ability to train only continuous prompts while freezing the language model. This approach greatly reduces the training time and storage for downstream tasks. In this work, we delve into the hierarchical relationship between the prompts and downstream text inputs. In prompt learning, the prefix prompt acts as a module to guide the downstream language model, establishing a hierarchical relationship between the prefix prompt and subsequent inputs. Furthermore, we explore the benefits of leveraging hyperbolic space for modeling hierarchical structures. We project representations of pre-trained models from Euclidean space into hyperbolic space using the Poincar{\'e} disk which effectively captures the hierarchical relationship between the prompt and input text. The experiments on natural language understanding (NLU) tasks illustrate that hyperbolic space can model the hierarchical relationship between prompt and text input. We release our code at \url{https://github.com/myaxxxxx/Hyperbolic-Prompt-Learning}.

Hyperbolic Representations for Prompt Learning

lrec-main.744

Poster

https://aclanthology.org/2024.lrec-main.745.bib

https://aclanthology.org/2024.lrec-main.745/

@inproceedings{zheng-etal-2024-hypergraph, title = "Hypergraph-Based Session Modeling: A Multi-Collaborative Self-Supervised Approach for Enhanced Recommender Systems", author = "Zheng, Xiangping and Wu, Bo and Zhang, Alex X. and Li, Wei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.745", pages = "8493--8504", abstract = "Session-based recommendation (SBR) is a challenging task that involves predicting a user{'}s next item click based on their recent session history. Presently, many state-of-the-art methodologies employ graph neural networks to model item transitions. Notwithstanding their impressive performance, graph-based models encounter significant challenges when confronted with intricate session dependencies and data sparsity in real-world scenarios, ultimately constraining their capacity to enhance recommendation accuracy. In recognition of these challenges, we introduce an innovative methodology known as {`}Mssen,{'} which stands for Multi-collaborative self-supervised learning in hypergraph neural networks. Mssen is meticulously crafted to adeptly discern user intent. Our approach initiates by representing session-based data as a hypergraph, adeptly capturing intricate, high-order relationships. Subsequently, we employ self-supervised learning on item-session hypergraphs to mitigate the challenges of data sparsity, all without necessitating manual fine-tuning, extensive search, or domain-specific expertise in augmentation selection. Comprehensive experimental analyses conducted across multiple datasets consistently underscore the superior performance of our approach when compared to existing methodologies.", }

Session-based recommendation (SBR) is a challenging task that involves predicting a user{'}s next item click based on their recent session history. Presently, many state-of-the-art methodologies employ graph neural networks to model item transitions. Notwithstanding their impressive performance, graph-based models encounter significant challenges when confronted with intricate session dependencies and data sparsity in real-world scenarios, ultimately constraining their capacity to enhance recommendation accuracy. In recognition of these challenges, we introduce an innovative methodology known as {`}Mssen,{'} which stands for Multi-collaborative self-supervised learning in hypergraph neural networks. Mssen is meticulously crafted to adeptly discern user intent. Our approach initiates by representing session-based data as a hypergraph, adeptly capturing intricate, high-order relationships. Subsequently, we employ self-supervised learning on item-session hypergraphs to mitigate the challenges of data sparsity, all without necessitating manual fine-tuning, extensive search, or domain-specific expertise in augmentation selection. Comprehensive experimental analyses conducted across multiple datasets consistently underscore the superior performance of our approach when compared to existing methodologies.

Hypergraph-Based Session Modeling: A Multi-Collaborative Self-Supervised Approach for Enhanced Recommender Systems

lrec-main.745

Poster

https://aclanthology.org/2024.lrec-main.746.bib

https://aclanthology.org/2024.lrec-main.746/

@inproceedings{wang-etal-2024-hypermr, title = "{H}yper{MR}: Hyperbolic Hypergraph Multi-hop Reasoning for Knowledge-based Visual Question Answering", author = "Wang, Bin and Xu, Fuyong and Liu, Peiyu and Zhu, Zhenfang", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.746", pages = "8505--8515", abstract = "Knowledge-based Visual Question Answering (KBVQA) is a challenging task, which aims to answer an image related question based on external knowledge. Most of the works describe the semantic distance using the actual Euclidean distance between two nodes, which leads to distortion in modeling knowledge graphs with hierarchical and scale-free structure in KBVQA, and limits the multi-hop reasoning capability of the model. In contrast, the hyperbolic space shows exciting prospects for low-distortion embedding of graphs with hierarchical and free-scale structure. In addition, we map the different stages of reasoning into multiple adjustable hyperbolic spaces, achieving low-distortion, fine-grained reasoning. Extensive experiments on the KVQA, PQ and PQL datasets demonstrate the effectiveness of HyperMR for strong-hierarchy knowledge graphs.", }

Knowledge-based Visual Question Answering (KBVQA) is a challenging task, which aims to answer an image related question based on external knowledge. Most of the works describe the semantic distance using the actual Euclidean distance between two nodes, which leads to distortion in modeling knowledge graphs with hierarchical and scale-free structure in KBVQA, and limits the multi-hop reasoning capability of the model. In contrast, the hyperbolic space shows exciting prospects for low-distortion embedding of graphs with hierarchical and free-scale structure. In addition, we map the different stages of reasoning into multiple adjustable hyperbolic spaces, achieving low-distortion, fine-grained reasoning. Extensive experiments on the KVQA, PQ and PQL datasets demonstrate the effectiveness of HyperMR for strong-hierarchy knowledge graphs.

HyperMR: Hyperbolic Hypergraph Multi-hop Reasoning for Knowledge-based Visual Question Answering

lrec-main.746

Poster

https://aclanthology.org/2024.lrec-main.747.bib

https://aclanthology.org/2024.lrec-main.747/

@inproceedings{li-etal-2024-hypertts, title = "{HYPERTTS}: Parameter Efficient Adaptation in Text to Speech Using Hypernetworks", author = "Li, Yingting and Bhardwaj, Rishabh and Mehrish, Ambuj and Cheng, Bo and Poria, Soujanya", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.747", pages = "8516--8527", abstract = "Neural speech synthesis, or text-to-speech (TTS), aims to transform a signal from the text domain to the speech domain. While developing TTS architectures that train and test on the same set of speakers has seen significant improvements, out-of-domain speaker performance still faces enormous limitations. Domain adaptation on a new set of speakers can be achieved by fine-tuning the whole model for each new domain, thus making it parameter-inefficient. This problem can be solved by Adapters that provide a parameter-efficient alternative to domain adaptation. Although famous in NLP, speech synthesis has not seen much improvement from Adapters. In this work, we present \textbf{HyperTTS}, which comprises a small learnable network, {``}hypernetwork{''}, that generates parameters of the Adapter blocks, allowing us to condition Adapters on speaker representations and making them dynamic. Extensive evaluations of two domain adaptation settings demonstrate its effectiveness in achieving state-of-the-art performance in the parameter-efficient regime. We also compare different variants of , comparing them with baselines in different studies. Promising results on the dynamic adaptation of adapter parameters using hypernetworks open up new avenues for domain-generic multi-speaker TTS systems. The audio samples and code are available at \url{https://github.com/declare-lab/HyperTTS}.", }

Neural speech synthesis, or text-to-speech (TTS), aims to transform a signal from the text domain to the speech domain. While developing TTS architectures that train and test on the same set of speakers has seen significant improvements, out-of-domain speaker performance still faces enormous limitations. Domain adaptation on a new set of speakers can be achieved by fine-tuning the whole model for each new domain, thus making it parameter-inefficient. This problem can be solved by Adapters that provide a parameter-efficient alternative to domain adaptation. Although famous in NLP, speech synthesis has not seen much improvement from Adapters. In this work, we present \textbf{HyperTTS}, which comprises a small learnable network, {``}hypernetwork{''}, that generates parameters of the Adapter blocks, allowing us to condition Adapters on speaker representations and making them dynamic. Extensive evaluations of two domain adaptation settings demonstrate its effectiveness in achieving state-of-the-art performance in the parameter-efficient regime. We also compare different variants of , comparing them with baselines in different studies. Promising results on the dynamic adaptation of adapter parameters using hypernetworks open up new avenues for domain-generic multi-speaker TTS systems. The audio samples and code are available at \url{https://github.com/declare-lab/HyperTTS}.

HYPERTTS: Parameter Efficient Adaptation in Text to Speech Using Hypernetworks

lrec-main.747

Poster

2404.04645

https://aclanthology.org/2024.lrec-main.748.bib

https://aclanthology.org/2024.lrec-main.748/

@inproceedings{lu-etal-2024-hyrr, title = "{HYRR}: Hybrid Infused Reranking for Passage Retrieval", author = "Lu, Jing and Hall, Keith and Ma, Ji and Ni, Jianmo", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.748", pages = "8528--8534", abstract = "Existing passage retrieval systems typically adopt a two-stage retrieve-then-rerank pipeline. To obtain an effective reranking model, many prior works have focused on improving the model architectures, such as leveraging powerful pretrained large language models (LLM) and designing better objective functions. However, less attention has been paid to the issue of collecting high-quality training data. In this paper, we propose HYRR, a framework for training robust reranking models. Specifically, we propose a simple but effective approach to select training data using hybrid retrievers. Our experiments show that the rerankers trained with HYRR are robust to different first-stage retrievers. Moreover, evaluations using MS MARCO and BEIR data sets demonstrate our proposed framework effectively generalizes to both supervised and zero-shot retrieval settings.", }

Existing passage retrieval systems typically adopt a two-stage retrieve-then-rerank pipeline. To obtain an effective reranking model, many prior works have focused on improving the model architectures, such as leveraging powerful pretrained large language models (LLM) and designing better objective functions. However, less attention has been paid to the issue of collecting high-quality training data. In this paper, we propose HYRR, a framework for training robust reranking models. Specifically, we propose a simple but effective approach to select training data using hybrid retrievers. Our experiments show that the rerankers trained with HYRR are robust to different first-stage retrievers. Moreover, evaluations using MS MARCO and BEIR data sets demonstrate our proposed framework effectively generalizes to both supervised and zero-shot retrieval settings.

HYRR: Hybrid Infused Reranking for Passage Retrieval

lrec-main.748

Poster

2212.10528

https://aclanthology.org/2024.lrec-main.749.bib

https://aclanthology.org/2024.lrec-main.749/

@inproceedings{liu-etal-2024-iad, title = "{IAD}: In-Context Learning Ability Decoupler of Large Language Models in Meta-Training", author = "Liu, Yuhan and Chen, Xiuying and Xing, Gao and Zhang, Ji and Yan, Rui", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.749", pages = "8535--8545", abstract = "Large Language Models (LLMs) exhibit remarkable In-Context Learning (ICL) ability, where the model learns tasks from prompts consisting of input-output examples. However, the pre-training objectives of LLMs often misalign with ICL objectives. They{'}re mainly pre-trained with methods like masked language modeling and next-sentence prediction. On the other hand, ICL leverages example pairs to guide the model in generating task-aware responses such as text classification and question-answering tasks. The basic pre-training task-related capabilities can sometimes overshadow or conflict with task-specific subtleties required in ICL. To address this, we propose an In-context learning Ability Decoupler (IAD). The model aims to separate the ICL ability from the general ability of LLMs in the meta-training phase, where the ICL-related parameters are separately tuned to adapt for ICL tasks. Concretely, we first identify the parameters that are suitable for ICL by transference-driven gradient importance. We then propose a new max-margin loss to emphasize the separation of the general and ICL abilities. The loss is defined as the difference between the output of ICL and the original LLM, aiming to prevent the overconfidence of the LLM. By meta-training these ICL-related parameters with max-margin loss, we enable the model to learn and adapt to new tasks with limited data effectively. Experimental results show that IAD{'}s capability yields state-of-the-art performance on benchmark datasets by utilizing only 30{\%} of the model{'}s parameters. Ablation study and detailed analysis prove the separation of the two abilities.", }

Large Language Models (LLMs) exhibit remarkable In-Context Learning (ICL) ability, where the model learns tasks from prompts consisting of input-output examples. However, the pre-training objectives of LLMs often misalign with ICL objectives. They{'}re mainly pre-trained with methods like masked language modeling and next-sentence prediction. On the other hand, ICL leverages example pairs to guide the model in generating task-aware responses such as text classification and question-answering tasks. The basic pre-training task-related capabilities can sometimes overshadow or conflict with task-specific subtleties required in ICL. To address this, we propose an In-context learning Ability Decoupler (IAD). The model aims to separate the ICL ability from the general ability of LLMs in the meta-training phase, where the ICL-related parameters are separately tuned to adapt for ICL tasks. Concretely, we first identify the parameters that are suitable for ICL by transference-driven gradient importance. We then propose a new max-margin loss to emphasize the separation of the general and ICL abilities. The loss is defined as the difference between the output of ICL and the original LLM, aiming to prevent the overconfidence of the LLM. By meta-training these ICL-related parameters with max-margin loss, we enable the model to learn and adapt to new tasks with limited data effectively. Experimental results show that IAD{'}s capability yields state-of-the-art performance on benchmark datasets by utilizing only 30{\%} of the model{'}s parameters. Ablation study and detailed analysis prove the separation of the two abilities.

IAD: In-Context Learning Ability Decoupler of Large Language Models in Meta-Training

lrec-main.749

Poster

https://aclanthology.org/2024.lrec-main.750.bib

https://aclanthology.org/2024.lrec-main.750/

@inproceedings{ge-etal-2024-idc, title = "{IDC}: Boost Text-to-image Retrieval via Indirect and Direct Connections", author = "Ge, Guowei and Hao, Kuangrong and Hao, Lingguang", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.750", pages = "8546--8555", abstract = "The Dual Encoders (DE) framework maps image and text inputs into a coordinated representation space, and calculates their similarity directly. On the other hand, the Cross Attention (CA) framework performs modalities interactions after completing the feature embedding of images and text, and then outputs a similarity score. For scenarios with bulk query requests or large query sets, the latter is more accurate, but the former is faster. Therefore, this work finds a new way to improve the retrieval accuracy of the DE framework by borrowing the advantages of the CA framework. Drawing inspiration from image captioning, we introduce a text decoder in the model training stage to simulate the cross-modal interaction function, like the CA framework. The text decoder is eventually discarded, aligning our model with the DE framework. Finally, to ensure training stability and prevent overfitting, we modify the Self-Distillation from Last Mini-Batch and apply it to the retrieval areas. Extensive experiments conducted on the MSCOCO and Flickr30K datasets validate the effectiveness of our proposed methods. Notably, our model achieves competitive results compared to state-of-the-art approaches on the Flickr30K dataset.", }

The Dual Encoders (DE) framework maps image and text inputs into a coordinated representation space, and calculates their similarity directly. On the other hand, the Cross Attention (CA) framework performs modalities interactions after completing the feature embedding of images and text, and then outputs a similarity score. For scenarios with bulk query requests or large query sets, the latter is more accurate, but the former is faster. Therefore, this work finds a new way to improve the retrieval accuracy of the DE framework by borrowing the advantages of the CA framework. Drawing inspiration from image captioning, we introduce a text decoder in the model training stage to simulate the cross-modal interaction function, like the CA framework. The text decoder is eventually discarded, aligning our model with the DE framework. Finally, to ensure training stability and prevent overfitting, we modify the Self-Distillation from Last Mini-Batch and apply it to the retrieval areas. Extensive experiments conducted on the MSCOCO and Flickr30K datasets validate the effectiveness of our proposed methods. Notably, our model achieves competitive results compared to state-of-the-art approaches on the Flickr30K dataset.

IDC: Boost Text-to-image Retrieval via Indirect and Direct Connections

lrec-main.750

Poster

https://aclanthology.org/2024.lrec-main.751.bib

https://aclanthology.org/2024.lrec-main.751/

@inproceedings{wang-etal-2024-ideate, title = "{IDEATE}: Detecting {AI}-Generated Text Using Internal and External Factual Structures", author = "Wang, Quan and Zhang, Licheng and Guo, Zikang and Mao, Zhendong", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.751", pages = "8556--8568", abstract = "The effective detection of AI-generated text is a vital principle to ensure responsible use of large language models (LLMs). Previous studies mainly focused on discovering and utilizing internal evidences contained in the text itself to perform the detection, while ignoring external evidences implicated in an established knowledge graph (KG) which may also be key discriminative factors between AI-generated and human-written text. To address this deficiency, we propose IDEATE, a novel hierarchical graph network that utilizes both internal and external factual structures to detect AI-generated text. IDEATE consists of a mention-level subgraph at the bottom to describe internal factual structures of mentioned entities reflected in the input text, and an entity-level subgraph at the top to describe external factual structures of mentioned entities reflected in an external KG. Hierarchical graph convolution is then applied successively on the two subgraphs, through which the two types of factual structures will be embedded into the output and used for the final detection. Extensive experiments on four benchmarking datasets show that IDEATE consistently outperforms current state-of-the-art methods in detecting text generated by various LLMs, ranging from GPT-2 to the more powerful ChatGPT, verifying the necessity and superiority of introducing external evidences for AI-generated text detection.", }

The effective detection of AI-generated text is a vital principle to ensure responsible use of large language models (LLMs). Previous studies mainly focused on discovering and utilizing internal evidences contained in the text itself to perform the detection, while ignoring external evidences implicated in an established knowledge graph (KG) which may also be key discriminative factors between AI-generated and human-written text. To address this deficiency, we propose IDEATE, a novel hierarchical graph network that utilizes both internal and external factual structures to detect AI-generated text. IDEATE consists of a mention-level subgraph at the bottom to describe internal factual structures of mentioned entities reflected in the input text, and an entity-level subgraph at the top to describe external factual structures of mentioned entities reflected in an external KG. Hierarchical graph convolution is then applied successively on the two subgraphs, through which the two types of factual structures will be embedded into the output and used for the final detection. Extensive experiments on four benchmarking datasets show that IDEATE consistently outperforms current state-of-the-art methods in detecting text generated by various LLMs, ranging from GPT-2 to the more powerful ChatGPT, verifying the necessity and superiority of introducing external evidences for AI-generated text detection.

IDEATE: Detecting AI-Generated Text Using Internal and External Factual Structures

lrec-main.751

Poster

https://aclanthology.org/2024.lrec-main.752.bib

https://aclanthology.org/2024.lrec-main.752/

@inproceedings{prochnow-etal-2024-idem, title = "{IDEM}: The {ID}ioms with {EM}otions Dataset for Emotion Recognition", author = {Prochnow, Alexander and Bendler, Johannes E. and Lange, Caroline and Tzavellos, Foivos Ioannis and G{\"o}ritzer, Bas Marco and ten Thij, Marijn and Batista-Navarro, Riza}, editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.752", pages = "8569--8579", abstract = "Idiomatic expressions are used in everyday language and typically convey affect, i.e., emotion. However, very little work investigating the extent to which automated methods can recognise emotions expressed in idiom-containing text has been undertaken. This can be attributed to the lack of emotion-labelled datasets that support the development and evaluation of such methods. In this paper, we present the IDioms with EMotions (IDEM) dataset consisting of a total of 9685 idiom-containing sentences that were generated and labelled with any one of 36 emotion types, with the help of the GPT-4 generative language model. Human validation by two independent annotators showed that more than 51{\%} of the generated sentences are ideal examples, with the annotators reaching an agreement rate of 62{\%} measured in terms of Cohen{'}s Kappa coefficient. To establish baseline performance on IDEM, various transformer-based emotion recognition approaches were implemented and evaluated. Results show that a RoBERTa model fine-tuned as a sequence classifier obtains a weighted F1-score of 58.73{\%}, when the sequence provided as input specifies the idiom contained in a given sentence, together with its definition. Since this input configuration is based on the assumption that the idiom contained in the given sentence is already known, we also sought to assess the feasibility of automatically identifying the idioms contained in IDEM sentences. To this end, a hybrid idiom identification approach combining a rule-based method and a deep learning-based model was developed, whose performance on IDEM was determined to be 84.99{\%} in terms of F1-score.", }

Idiomatic expressions are used in everyday language and typically convey affect, i.e., emotion. However, very little work investigating the extent to which automated methods can recognise emotions expressed in idiom-containing text has been undertaken. This can be attributed to the lack of emotion-labelled datasets that support the development and evaluation of such methods. In this paper, we present the IDioms with EMotions (IDEM) dataset consisting of a total of 9685 idiom-containing sentences that were generated and labelled with any one of 36 emotion types, with the help of the GPT-4 generative language model. Human validation by two independent annotators showed that more than 51{\%} of the generated sentences are ideal examples, with the annotators reaching an agreement rate of 62{\%} measured in terms of Cohen{'}s Kappa coefficient. To establish baseline performance on IDEM, various transformer-based emotion recognition approaches were implemented and evaluated. Results show that a RoBERTa model fine-tuned as a sequence classifier obtains a weighted F1-score of 58.73{\%}, when the sequence provided as input specifies the idiom contained in a given sentence, together with its definition. Since this input configuration is based on the assumption that the idiom contained in the given sentence is already known, we also sought to assess the feasibility of automatically identifying the idioms contained in IDEM sentences. To this end, a hybrid idiom identification approach combining a rule-based method and a deep learning-based model was developed, whose performance on IDEM was determined to be 84.99{\%} in terms of F1-score.

IDEM: The IDioms with EMotions Dataset for Emotion Recognition

lrec-main.752

Poster

https://aclanthology.org/2024.lrec-main.753.bib

https://aclanthology.org/2024.lrec-main.753/

@inproceedings{hughes-song-2024-identifying, title = "Identifying and Aligning Medical Claims Made on Social Media with Medical Evidence", author = "Hughes, Anthony James and Song, Xingyi", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.753", pages = "8580--8593", abstract = "Evidence-based medicine is the practise of making medical decisions that adhere to the latest, and best known evidence at that time. Currently, the best evidence is often found in the form of documents, such as randomized control trials, meta-analyses and systematic reviews. This research focuses on aligning medical claims made on social media platforms with this medical evidence. By doing so, individuals without medical expertise can more effectively assess the veracity of such medical claims. We study three core tasks: identifying medical claims, extracting medical vocabulary from these claims, and retrieving evidence relevant to those identified medical claims. We propose a novel system that can generate synthetic medical claims to aid each of these core tasks. We additionally introduce a novel dataset produced by our synthetic generator that, when applied to these tasks, demonstrates not only a more flexible and holistic approach, but also an improvement in all comparable metrics. We make our dataset, the Expansive Medical Claim Corpus (EMCC), available at https://zenodo.org/records/8321460.", }

Evidence-based medicine is the practise of making medical decisions that adhere to the latest, and best known evidence at that time. Currently, the best evidence is often found in the form of documents, such as randomized control trials, meta-analyses and systematic reviews. This research focuses on aligning medical claims made on social media platforms with this medical evidence. By doing so, individuals without medical expertise can more effectively assess the veracity of such medical claims. We study three core tasks: identifying medical claims, extracting medical vocabulary from these claims, and retrieving evidence relevant to those identified medical claims. We propose a novel system that can generate synthetic medical claims to aid each of these core tasks. We additionally introduce a novel dataset produced by our synthetic generator that, when applied to these tasks, demonstrates not only a more flexible and holistic approach, but also an improvement in all comparable metrics. We make our dataset, the Expansive Medical Claim Corpus (EMCC), available at https://zenodo.org/records/8321460.

Identifying and Aligning Medical Claims Made on Social Media with Medical Evidence

lrec-main.753

Poster

2405.11219

https://aclanthology.org/2024.lrec-main.754.bib

https://aclanthology.org/2024.lrec-main.754/

@inproceedings{mendes-caseli-2024-identifying, title = "Identifying Fine-grained Depression Signs in Social Media Posts", author = "Mendes, Augusto R. and Caseli, Helena", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.754", pages = "8594--8604", abstract = "Natural Language Processing has already proven to be an effective tool for helping in the identification of mental health disorders in text. However, most studies limit themselves to a binary classification setup or base their label set on pre-established resources. By doing so, they don{'}t explicitly model many common ways users can express their depression online, limiting our understanding of what kind of depression signs such models can accurately classify. This study evaluates how machine learning techniques deal with the classification of a fine-grained set of 21 depression signs in social media posts from Brazilian undergraduate students. We found out that model performance is not necessarily driven by a depression sign{'}s frequency on social media posts, since evaluated machine learning techniques struggle to classify the majority of signs of depression typically present in posts. Thus, model performance seems to be more related to the inherent difficulty of identifying a given sign than with its occurrence frequency.", }

Natural Language Processing has already proven to be an effective tool for helping in the identification of mental health disorders in text. However, most studies limit themselves to a binary classification setup or base their label set on pre-established resources. By doing so, they don{'}t explicitly model many common ways users can express their depression online, limiting our understanding of what kind of depression signs such models can accurately classify. This study evaluates how machine learning techniques deal with the classification of a fine-grained set of 21 depression signs in social media posts from Brazilian undergraduate students. We found out that model performance is not necessarily driven by a depression sign{'}s frequency on social media posts, since evaluated machine learning techniques struggle to classify the majority of signs of depression typically present in posts. Thus, model performance seems to be more related to the inherent difficulty of identifying a given sign than with its occurrence frequency.

Identifying Fine-grained Depression Signs in Social Media Posts

lrec-main.754

Poster

https://aclanthology.org/2024.lrec-main.755.bib

https://aclanthology.org/2024.lrec-main.755/

@inproceedings{sakaguchi-etal-2024-identifying, title = "Identifying Source Language Expressions for Pre-editing in Machine Translation", author = "Sakaguchi, Norizo and Murawaki, Yugo and Chu, Chenhui and Kurohashi, Sadao", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.755", pages = "8605--8616", abstract = "Machine translation-mediated communication can benefit from pre-editing source language texts to ensure accurate transmission of intended meaning in the target language. The primary challenge lies in identifying source language expressions that pose difficulties in translation. In this paper, we hypothesize that such expressions tend to be distinctive features of texts originally written in the source language (native language) rather than translations generated from the target language into the source language (machine translation). To identify such expressions, we train a neural classifier to distinguish native language from machine translation, and subsequently isolate the expressions that contribute to the model{'}s prediction of native language. Our manual evaluation revealed that our method successfully identified characteristic expressions of the native language, despite the noise and the inherent nuances of the task. We also present case studies where we edit the identified expressions to improve translation quality.", }

Machine translation-mediated communication can benefit from pre-editing source language texts to ensure accurate transmission of intended meaning in the target language. The primary challenge lies in identifying source language expressions that pose difficulties in translation. In this paper, we hypothesize that such expressions tend to be distinctive features of texts originally written in the source language (native language) rather than translations generated from the target language into the source language (machine translation). To identify such expressions, we train a neural classifier to distinguish native language from machine translation, and subsequently isolate the expressions that contribute to the model{'}s prediction of native language. Our manual evaluation revealed that our method successfully identified characteristic expressions of the native language, despite the noise and the inherent nuances of the task. We also present case studies where we edit the identified expressions to improve translation quality.

Identifying Source Language Expressions for Pre-editing in Machine Translation

lrec-main.755

Poster

https://aclanthology.org/2024.lrec-main.756.bib

https://aclanthology.org/2024.lrec-main.756/

@inproceedings{reimerink-etal-2024-ideological, title = "Ideological Knowledge Representation: Framing Climate Change in {E}co{L}exicon", author = "Reimerink, Arianne and Cabezas-Garc{\'\i}a, Melania and Le{\'o}n-Ara{\'u}z, Pilar and Faber, Pamela", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.756", pages = "8617--8626", abstract = "Culture is underrepresented in terminological resources and ideology is an especially complicated cultural aspect to convey. This complexity stems from the intertwined relationships among the discourse community of politicians, the media and the general public, as well as their interactions with scientific knowledge. Nevertheless, terminological resources should provide the necessary information to understand the political perspective taken in discourse on scientific issues with a high political profile. As in all specialized domains, environmental concepts and terms are subject to dynamism and variation (Le{\'o}n-Ara{\'u}z, 2017). Cognitive term variants (e.g., climate change, climate crisis) are of particular interest because of their presence in political discourse and their potential to influence climate actions. They can be used to reflect multidimensionality, imprecision or ideological attachment. This paper describes a method based on framing in Communication Studies to extract ideological knowledge from corpora. We used Spanish and English parliamentary debates (ParlaMint 2.1) and annotated the interventions that included a term variant of climate change according to an adapted version of the frames proposed by Bolsen and Shapiro (2018). The results showed how climate change discourse changes across de ideological spectrum and we give a proposal on how to represent that knowledge in an environmental TKB on the environment.", }

Culture is underrepresented in terminological resources and ideology is an especially complicated cultural aspect to convey. This complexity stems from the intertwined relationships among the discourse community of politicians, the media and the general public, as well as their interactions with scientific knowledge. Nevertheless, terminological resources should provide the necessary information to understand the political perspective taken in discourse on scientific issues with a high political profile. As in all specialized domains, environmental concepts and terms are subject to dynamism and variation (Le{\'o}n-Ara{\'u}z, 2017). Cognitive term variants (e.g., climate change, climate crisis) are of particular interest because of their presence in political discourse and their potential to influence climate actions. They can be used to reflect multidimensionality, imprecision or ideological attachment. This paper describes a method based on framing in Communication Studies to extract ideological knowledge from corpora. We used Spanish and English parliamentary debates (ParlaMint 2.1) and annotated the interventions that included a term variant of climate change according to an adapted version of the frames proposed by Bolsen and Shapiro (2018). The results showed how climate change discourse changes across de ideological spectrum and we give a proposal on how to represent that knowledge in an environmental TKB on the environment.

Ideological Knowledge Representation: Framing Climate Change in EcoLexicon

lrec-main.756

Poster

https://aclanthology.org/2024.lrec-main.757.bib

https://aclanthology.org/2024.lrec-main.757/

@inproceedings{ghosh-roy-han-2024-ilciter, title = "{ILC}ite{R}: Evidence-grounded Interpretable Local Citation Recommendation", author = "Ghosh Roy, Sayar and Han, Jiawei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.757", pages = "8627--8638", abstract = "Existing Machine Learning approaches for local citation recommendation directly map or translate a query, which is typically a claim or an entity mention, to citation-worthy research papers. Within such a formulation, it is challenging to pinpoint why one should cite a specific research paper for a particular query, leading to limited recommendation interpretability. To alleviate this, we introduce the evidence-grounded local citation recommendation task, where the target latent space comprises evidence spans for recommending specific papers. Using a distantly-supervised evidence retrieval and multi-step re-ranking framework, our proposed system, ILCiteR, recommends papers to cite for a query grounded on similar evidence spans extracted from the existing research literature. Unlike past formulations that simply output recommendations, ILCiteR retrieves ranked lists of evidence span and recommended paper pairs. Secondly, previously proposed neural models for citation recommendation require expensive training on massive labeled data, ideally after every significant update to the pool of candidate papers. In contrast, ILCiteR relies solely on distant supervision from a dynamic evidence database and pre-trained Transformer-based Language Models without any model training. We contribute a novel dataset for the evidence-grounded local citation recommendation task and demonstrate the efficacy of our proposed conditional neural rank-ensembling approach for re-ranking evidence spans.", }

Existing Machine Learning approaches for local citation recommendation directly map or translate a query, which is typically a claim or an entity mention, to citation-worthy research papers. Within such a formulation, it is challenging to pinpoint why one should cite a specific research paper for a particular query, leading to limited recommendation interpretability. To alleviate this, we introduce the evidence-grounded local citation recommendation task, where the target latent space comprises evidence spans for recommending specific papers. Using a distantly-supervised evidence retrieval and multi-step re-ranking framework, our proposed system, ILCiteR, recommends papers to cite for a query grounded on similar evidence spans extracted from the existing research literature. Unlike past formulations that simply output recommendations, ILCiteR retrieves ranked lists of evidence span and recommended paper pairs. Secondly, previously proposed neural models for citation recommendation require expensive training on massive labeled data, ideally after every significant update to the pool of candidate papers. In contrast, ILCiteR relies solely on distant supervision from a dynamic evidence database and pre-trained Transformer-based Language Models without any model training. We contribute a novel dataset for the evidence-grounded local citation recommendation task and demonstrate the efficacy of our proposed conditional neural rank-ensembling approach for re-ranking evidence spans.

ILCiteR: Evidence-grounded Interpretable Local Citation Recommendation

lrec-main.757

Poster

2403.08737

https://aclanthology.org/2024.lrec-main.758.bib

https://aclanthology.org/2024.lrec-main.758/

@inproceedings{mirza-etal-2024-illuminer, title = "{ILLUMINER}: Instruction-tuned Large Language Models as Few-shot Intent Classifier and Slot Filler", author = "Mirza, Paramita and Sudhi, Viju and Sahoo, Soumya Ranjan and Bhat, Sinchana Ramakanth", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.758", pages = "8639--8651", abstract = "State-of-the-art intent classification (IC) and slot filling (SF) methods often rely on data-intensive deep learning models, limiting their practicality for industry applications. Large language models on the other hand, particularly instruction-tuned models (Instruct-LLMs), exhibit remarkable zero-shot performance across various natural language tasks. This study evaluates Instruct-LLMs on popular benchmark datasets for IC and SF, emphasizing their capacity to learn from fewer examples. We introduce ILLUMINER, an approach framing IC and SF as language generation tasks for Instruct-LLMs, with a more efficient SF-prompting method compared to prior work. A comprehensive comparison with multiple baselines shows that our approach, using the FLAN-T5 11B model, outperforms the state-of-the-art joint IC+SF method and in-context learning with GPT3.5 (175B), particularly in slot filling by 11.1{--}32.2 percentage points. Additionally, our in-depth ablation study demonstrates that parameter-efficient fine-tuning requires less than 6{\%} of training data to yield comparable performance with traditional full-weight fine-tuning.", }

State-of-the-art intent classification (IC) and slot filling (SF) methods often rely on data-intensive deep learning models, limiting their practicality for industry applications. Large language models on the other hand, particularly instruction-tuned models (Instruct-LLMs), exhibit remarkable zero-shot performance across various natural language tasks. This study evaluates Instruct-LLMs on popular benchmark datasets for IC and SF, emphasizing their capacity to learn from fewer examples. We introduce ILLUMINER, an approach framing IC and SF as language generation tasks for Instruct-LLMs, with a more efficient SF-prompting method compared to prior work. A comprehensive comparison with multiple baselines shows that our approach, using the FLAN-T5 11B model, outperforms the state-of-the-art joint IC+SF method and in-context learning with GPT3.5 (175B), particularly in slot filling by 11.1{--}32.2 percentage points. Additionally, our in-depth ablation study demonstrates that parameter-efficient fine-tuning requires less than 6{\%} of training data to yield comparable performance with traditional full-weight fine-tuning.

ILLUMINER: Instruction-tuned Large Language Models as Few-shot Intent Classifier and Slot Filler

lrec-main.758

Poster

2403.17536

https://huggingface.co/papers/2403.17536

https://aclanthology.org/2024.lrec-main.759.bib

https://aclanthology.org/2024.lrec-main.759/

@inproceedings{zhang-wan-2024-image, title = "Image Matters: A New Dataset and Empirical Study for Multimodal Hyperbole Detection", author = "Zhang, Huixuan and Wan, Xiaojun", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.759", pages = "8652--8661", abstract = "Hyperbole, or exaggeration, is a common linguistic phenomenon. The detection of hyperbole is an important part of understanding human expression. There have been several studies on hyperbole detection, but most of which focus on text modality only. However, with the development of social media, people can create hyperbolic expressions with various modalities, including text, images, videos, etc. In this paper, we focus on multimodal hyperbole detection. We create a multimodal detection dataset from Weibo (a Chinese social media) and carry out some studies on it. We treat the text and image from a piece of weibo as two modalities and explore the role of text and image for hyperbole detection. Different pre-trained multimodal encoders are also evaluated on this downstream task to show their performance. Besides, since this dataset is constructed from five different keywords, we also evaluate the cross-domain performance of different models. These studies can serve as a benchmark and point out the direction of further study on multimodal hyperbole detection.", }

Hyperbole, or exaggeration, is a common linguistic phenomenon. The detection of hyperbole is an important part of understanding human expression. There have been several studies on hyperbole detection, but most of which focus on text modality only. However, with the development of social media, people can create hyperbolic expressions with various modalities, including text, images, videos, etc. In this paper, we focus on multimodal hyperbole detection. We create a multimodal detection dataset from Weibo (a Chinese social media) and carry out some studies on it. We treat the text and image from a piece of weibo as two modalities and explore the role of text and image for hyperbole detection. Different pre-trained multimodal encoders are also evaluated on this downstream task to show their performance. Besides, since this dataset is constructed from five different keywords, we also evaluate the cross-domain performance of different models. These studies can serve as a benchmark and point out the direction of further study on multimodal hyperbole detection.

Image Matters: A New Dataset and Empirical Study for Multimodal Hyperbole Detection

lrec-main.759

Poster

2307.00209

https://aclanthology.org/2024.lrec-main.760.bib

https://aclanthology.org/2024.lrec-main.760/

@inproceedings{lee-bloem-2024-impact, title = "Impact of Task Adapting on Transformer Models for Targeted Sentiment Analysis in {C}roatian Headlines", author = "Lee, Sofia and Bloem, Jelke", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.760", pages = "8662--8674", abstract = "Transformer models, such as BERT, are often taken off-the-shelf and then fine-tuned on a downstream task. Although this is sufficient for many tasks, low-resource settings require special attention. We demonstrate an approach of performing an extra stage of self-supervised task-adaptive pre-training to a number of Croatian-supporting Transformer models. In particular, we focus on approaches to language, domain, and task adaptation. The task in question is targeted sentiment analysis for Croatian news headlines. We produce new state-of-the-art results (F1 = 0.781), but the highest performing model still struggles with irony and implicature. Overall, we find that task-adaptive pre-training benefits massively multilingual models but not Croatian-dominant models.", }

Transformer models, such as BERT, are often taken off-the-shelf and then fine-tuned on a downstream task. Although this is sufficient for many tasks, low-resource settings require special attention. We demonstrate an approach of performing an extra stage of self-supervised task-adaptive pre-training to a number of Croatian-supporting Transformer models. In particular, we focus on approaches to language, domain, and task adaptation. The task in question is targeted sentiment analysis for Croatian news headlines. We produce new state-of-the-art results (F1 = 0.781), but the highest performing model still struggles with irony and implicature. Overall, we find that task-adaptive pre-training benefits massively multilingual models but not Croatian-dominant models.

Impact of Task Adapting on Transformer Models for Targeted Sentiment Analysis in Croatian Headlines

lrec-main.760

Poster

https://aclanthology.org/2024.lrec-main.761.bib

https://aclanthology.org/2024.lrec-main.761/

@inproceedings{cercas-curry-etal-2024-impoverished, title = "Impoverished Language Technology: The Lack of (Social) Class in {NLP}", author = "Cercas Curry, Amanda and Talat, Zeerak and Hovy, Dirk", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.761", pages = "8675--8682", abstract = "Since Labov{'}s foundational 1964 work on the social stratification of language, linguistics has dedicated concerted efforts towards understanding the relationships between socio-demographic factors and language production and perception. Despite the large body of evidence identifying significant relationships between socio-demographic factors and language production, relatively few of these factors have been investigated in the context of NLP technology. While age and gender are well covered, Labov{'}s initial target, socio-economic class, is largely absent. We survey the existing Natural Language Processing (NLP) literature and find that only 20 papers even mention socio-economic status. However, the majority of those papers do not engage with class beyond collecting information of annotator-demographics. Given this research lacuna, we provide a definition of class that can be operationalised by NLP researchers, and argue for including socio-economic class in future language technologies.", }

Since Labov{'}s foundational 1964 work on the social stratification of language, linguistics has dedicated concerted efforts towards understanding the relationships between socio-demographic factors and language production and perception. Despite the large body of evidence identifying significant relationships between socio-demographic factors and language production, relatively few of these factors have been investigated in the context of NLP technology. While age and gender are well covered, Labov{'}s initial target, socio-economic class, is largely absent. We survey the existing Natural Language Processing (NLP) literature and find that only 20 papers even mention socio-economic status. However, the majority of those papers do not engage with class beyond collecting information of annotator-demographics. Given this research lacuna, we provide a definition of class that can be operationalised by NLP researchers, and argue for including socio-economic class in future language technologies.

Impoverished Language Technology: The Lack of (Social) Class in NLP

lrec-main.761

Poster

2403.03874

https://aclanthology.org/2024.lrec-main.762.bib

https://aclanthology.org/2024.lrec-main.762/

@inproceedings{lu-etal-2024-improved, title = "Improved Neural Protoform Reconstruction via Reflex Prediction", author = "Lu, Liang and Wang, Jingzhi and Mortensen, David R.", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.762", pages = "8683--8707", abstract = "Protolanguage reconstruction is central to historical linguistics. The comparative method, one of the most influential theoretical and methodological frameworks in the history of the language sciences, allows linguists to infer protoforms (reconstructed ancestral words) from their reflexes (related modern words) based on the assumption of regular sound change. Not surprisingly, numerous computational linguists have attempted to operationalize comparative reconstruction through various computational models, the most successful of which have been supervised encoder-decoder models, which treat the problem of predicting protoforms given sets of reflexes as a sequence-to-sequence problem. We argue that this framework ignores one of the most important aspects of the comparative method: not only should protoforms be inferable from cognate sets (sets of related reflexes) but the reflexes should also be inferable from the protoforms. Leveraging another line of research{---}reflex prediction{---}we propose a system in which candidate protoforms from a reconstruction model are reranked by a reflex prediction model. We show that this more complete implementation of the comparative method allows us to surpass state-of-the-art protoform reconstruction methods on three of four Chinese and Romance datasets.", }

Protolanguage reconstruction is central to historical linguistics. The comparative method, one of the most influential theoretical and methodological frameworks in the history of the language sciences, allows linguists to infer protoforms (reconstructed ancestral words) from their reflexes (related modern words) based on the assumption of regular sound change. Not surprisingly, numerous computational linguists have attempted to operationalize comparative reconstruction through various computational models, the most successful of which have been supervised encoder-decoder models, which treat the problem of predicting protoforms given sets of reflexes as a sequence-to-sequence problem. We argue that this framework ignores one of the most important aspects of the comparative method: not only should protoforms be inferable from cognate sets (sets of related reflexes) but the reflexes should also be inferable from the protoforms. Leveraging another line of research{---}reflex prediction{---}we propose a system in which candidate protoforms from a reconstruction model are reranked by a reflex prediction model. We show that this more complete implementation of the comparative method allows us to surpass state-of-the-art protoform reconstruction methods on three of four Chinese and Romance datasets.

Improved Neural Protoform Reconstruction via Reflex Prediction

lrec-main.762

Poster

2403.18769

https://huggingface.co/papers/2403.18769

https://aclanthology.org/2024.lrec-main.763.bib

https://aclanthology.org/2024.lrec-main.763/

@inproceedings{zawbaa-etal-2024-improved, title = "Improved Out-of-Scope Intent Classification with Dual Encoding and Threshold-based Re-Classification", author = "Zawbaa, Hossam and Rashwan, Wael and Dutta, Sourav and Assem, Haytham", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.763", pages = "8708--8718", abstract = "Detecting out-of-scope user utterances is essential for task-oriented dialogues and intent classification. Current methodologies face difficulties with the unpredictable distribution of outliers and often rely on assumptions about data distributions. We present the Dual Encoder for Threshold-Based Re-Classification (DETER) to address these challenges. This end-to-end framework efficiently detects out-of-scope intents without requiring assumptions on data distributions or additional post-processing steps. The core of DETER utilizes dual text encoders, the Universal Sentence Encoder (USE) and the Transformer-based Denoising AutoEncoder (TSDAE), to generate user utterance embeddings, which are classified through a branched neural architecture. Further, DETER generates synthetic outliers using self-supervision and incorporates out-of-scope phrases from open-domain datasets. This approach ensures a comprehensive training set for out-of-scope detection. Additionally, a threshold-based re-classification mechanism refines the model{'}s initial predictions. Evaluations on the CLINC-150, Stackoverflow, and Banking77 datasets demonstrate DETER{'}s efficacy. Our model outperforms previous benchmarks, achieving an increase of up to 13{\%} and 5{\%} in F1 score for known and unknown intents on CLINC-150 and Stackoverflow, and 16{\%} for known and 24{\%} for unknown intents on Banking77. The source code has been released at https://github.com/Hossam-Mohammed-tech/Intent{\_}Classification{\_}OOS.", }

Detecting out-of-scope user utterances is essential for task-oriented dialogues and intent classification. Current methodologies face difficulties with the unpredictable distribution of outliers and often rely on assumptions about data distributions. We present the Dual Encoder for Threshold-Based Re-Classification (DETER) to address these challenges. This end-to-end framework efficiently detects out-of-scope intents without requiring assumptions on data distributions or additional post-processing steps. The core of DETER utilizes dual text encoders, the Universal Sentence Encoder (USE) and the Transformer-based Denoising AutoEncoder (TSDAE), to generate user utterance embeddings, which are classified through a branched neural architecture. Further, DETER generates synthetic outliers using self-supervision and incorporates out-of-scope phrases from open-domain datasets. This approach ensures a comprehensive training set for out-of-scope detection. Additionally, a threshold-based re-classification mechanism refines the model{'}s initial predictions. Evaluations on the CLINC-150, Stackoverflow, and Banking77 datasets demonstrate DETER{'}s efficacy. Our model outperforms previous benchmarks, achieving an increase of up to 13{\%} and 5{\%} in F1 score for known and unknown intents on CLINC-150 and Stackoverflow, and 16{\%} for known and 24{\%} for unknown intents on Banking77. The source code has been released at https://github.com/Hossam-Mohammed-tech/Intent{\_}Classification{\_}OOS.

Improved Out-of-Scope Intent Classification with Dual Encoding and Threshold-based Re-Classification

lrec-main.763

Poster

2405.19967

https://aclanthology.org/2024.lrec-main.764.bib

https://aclanthology.org/2024.lrec-main.764/

@inproceedings{shahriar-barbosa-2024-improving, title = "Improving {B}engali and {H}indi Large Language Models", author = "Shahriar, Arif and Barbosa, Denilson", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.764", pages = "8719--8731", abstract = "Despite being widely spoken worldwide, Bengali and Hindi are low-resource languages. The state-of-the-art in modeling such languages uses BERT and the Wordpiece tokenizer. We observed that the Wordpiece tokenizer often breaks words into meaningless tokens, failing to separate roots from affixes. Moreover, Wordpiece does not take into account fine-grained character-level information. We hypothesize that modeling fine-grained character-level information or interactions between roots and affixes helps with modeling highly inflected and morphologically complex languages such as Bengali and Hindi. We used BERT with two different tokenizers - a Unigram tokenizer and a character-level tokenizer and observed better performance. Then, we pretrained four language models accordingly - Bengali Unigram BERT, Hindi Unigram BERT, Bengali Character BERT, and Hindi Character BERT, and evaluated them for masked token detection, both in correct and erroneous settings, across many NLU tasks. We provide experimental evidence that Unigram and character-level tokenizers lead to better pretrained models for Bengali and Hindi, outperforming the previous state-of-the-art and BERT with Wordpiece vocabulary. We conduct the first study investigating the efficacy of different tokenization methods in modeling Bengali and Hindi.", }

Despite being widely spoken worldwide, Bengali and Hindi are low-resource languages. The state-of-the-art in modeling such languages uses BERT and the Wordpiece tokenizer. We observed that the Wordpiece tokenizer often breaks words into meaningless tokens, failing to separate roots from affixes. Moreover, Wordpiece does not take into account fine-grained character-level information. We hypothesize that modeling fine-grained character-level information or interactions between roots and affixes helps with modeling highly inflected and morphologically complex languages such as Bengali and Hindi. We used BERT with two different tokenizers - a Unigram tokenizer and a character-level tokenizer and observed better performance. Then, we pretrained four language models accordingly - Bengali Unigram BERT, Hindi Unigram BERT, Bengali Character BERT, and Hindi Character BERT, and evaluated them for masked token detection, both in correct and erroneous settings, across many NLU tasks. We provide experimental evidence that Unigram and character-level tokenizers lead to better pretrained models for Bengali and Hindi, outperforming the previous state-of-the-art and BERT with Wordpiece vocabulary. We conduct the first study investigating the efficacy of different tokenization methods in modeling Bengali and Hindi.

Improving Bengali and Hindi Large Language Models

lrec-main.764

Poster

https://aclanthology.org/2024.lrec-main.765.bib

https://aclanthology.org/2024.lrec-main.765/

@inproceedings{dou-etal-2024-improving, title = "Improving {C}hinese Named Entity Recognition with Multi-grained Words and Part-of-Speech Tags via Joint Modeling", author = "Dou, Chenhui and Gong, Chen and Li, Zhenghua and Wang, Zhefeng and Huai, Baoxing and Zhang, Min", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.765", pages = "8732--8742", abstract = "Nowadays, character-based sequence labeling becomes the mainstream Chinese named entity recognition (CNER) approach, instead of word-based methods, since the latter degrades performance due to propagation of word segmentation (WS) errors. To make use of WS information, previous studies usually learn CNER and WS simultaneously with multi-task learning (MTL) framework, or treat WS information as extra guide features for CNER model, in which the utilization of WS information is indirect and shallow. In light of the complementary information inside multi-grained words, and the close connection between named entities and part-of-speech (POS) tags, this work proposes a tree parsing approach for joint modeling CNER, multi-grained word segmentation (MWS) and POS tagging tasks simultaneously. Specifically, we first propose a unified tree representation for MWS, POS tagging, and CNER.Then, we automatically construct the MWS-POS-NER data based on the unified tree representation for model training. Finally, we present a two-stage joint tree parsing framework. Experimental results on OntoNotes4 and OntoNotes5 show that our proposed approach of jointly modeling CNER with MWS and POS tagging achieves better or comparable performance with latest methods.", }

Nowadays, character-based sequence labeling becomes the mainstream Chinese named entity recognition (CNER) approach, instead of word-based methods, since the latter degrades performance due to propagation of word segmentation (WS) errors. To make use of WS information, previous studies usually learn CNER and WS simultaneously with multi-task learning (MTL) framework, or treat WS information as extra guide features for CNER model, in which the utilization of WS information is indirect and shallow. In light of the complementary information inside multi-grained words, and the close connection between named entities and part-of-speech (POS) tags, this work proposes a tree parsing approach for joint modeling CNER, multi-grained word segmentation (MWS) and POS tagging tasks simultaneously. Specifically, we first propose a unified tree representation for MWS, POS tagging, and CNER.Then, we automatically construct the MWS-POS-NER data based on the unified tree representation for model training. Finally, we present a two-stage joint tree parsing framework. Experimental results on OntoNotes4 and OntoNotes5 show that our proposed approach of jointly modeling CNER with MWS and POS tagging achieves better or comparable performance with latest methods.

Improving Chinese Named Entity Recognition with Multi-grained Words and Part-of-Speech Tags via Joint Modeling

lrec-main.765

Poster

https://aclanthology.org/2024.lrec-main.766.bib

https://aclanthology.org/2024.lrec-main.766/

@inproceedings{kim-etal-2024-improving, title = "Improving Content Recommendation: Knowledge Graph-Based Semantic Contrastive Learning for Diversity and Cold-Start Users", author = "Kim, Yejin and Rome, Scott and Foley, Kevin and Nankani, Mayur and Melamed, Rimon and Morales, Javier and Yadav, Abhay K. and Peifer, Maria and Hamidian, Sardar and Huang, H. Howie", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.766", pages = "8743--8755", abstract = "Addressing the challenges related to data sparsity, cold-start problems, and diversity in recommendation systems is both crucial and demanding. Many current solutions leverage knowledge graphs to tackle these issues by combining both item-based and user-item collaborative signals. A common trend in these approaches focuses on improving ranking performance at the cost of escalating model complexity, reducing diversity, and complicating the task. It is essential to provide recommendations that are both personalized and diverse, rather than solely relying on achieving high rank-based performance, such as Click-through rate, Recall, etc. In this paper, we propose a hybrid multi-task learning approach, training on user-item and item-item interactions. We apply item-based contrastive learning on descriptive text, sampling positive and negative pairs based on item metadata. Our approach allows the model to better understand the relationships between entities within the knowledge graph by utilizing semantic information from text. It leads to more accurate, relevant, and diverse user recommendations and a benefit that extends even to cold-start users who have few interactions with items. We perform extensive experiments on two widely used datasets to validate the effectiveness of our approach. Our findings demonstrate that jointly training user-item interactions and item-based signals using synopsis text is highly effective. Furthermore, our results provide evidence that item-based contrastive learning enhances the quality of entity embeddings, as indicated by metrics such as uniformity and alignment.", }

Addressing the challenges related to data sparsity, cold-start problems, and diversity in recommendation systems is both crucial and demanding. Many current solutions leverage knowledge graphs to tackle these issues by combining both item-based and user-item collaborative signals. A common trend in these approaches focuses on improving ranking performance at the cost of escalating model complexity, reducing diversity, and complicating the task. It is essential to provide recommendations that are both personalized and diverse, rather than solely relying on achieving high rank-based performance, such as Click-through rate, Recall, etc. In this paper, we propose a hybrid multi-task learning approach, training on user-item and item-item interactions. We apply item-based contrastive learning on descriptive text, sampling positive and negative pairs based on item metadata. Our approach allows the model to better understand the relationships between entities within the knowledge graph by utilizing semantic information from text. It leads to more accurate, relevant, and diverse user recommendations and a benefit that extends even to cold-start users who have few interactions with items. We perform extensive experiments on two widely used datasets to validate the effectiveness of our approach. Our findings demonstrate that jointly training user-item interactions and item-based signals using synopsis text is highly effective. Furthermore, our results provide evidence that item-based contrastive learning enhances the quality of entity embeddings, as indicated by metrics such as uniformity and alignment.

Improving Content Recommendation: Knowledge Graph-Based Semantic Contrastive Learning for Diversity and Cold-Start Users

lrec-main.766

Poster

2403.18667

https://aclanthology.org/2024.lrec-main.767.bib

https://aclanthology.org/2024.lrec-main.767/

@inproceedings{zhang-etal-2024-improving-continual, title = "Improving Continual Few-shot Relation Extraction through Relational Knowledge Distillation and Prototype Augmentation", author = "Zhang, Zhiheng and Zeng, Daojian and Bai, Xue", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.767", pages = "8756--8767", abstract = "In this paper, we focus on the challenging yet practical problem of Continual Few-shot Relation Extraction (CFRE), which involves extracting relations in the continuous and iterative arrival of new data with only a few labeled examples. The main challenges in CFRE are overfitting due to few-shot learning and catastrophic forgetting caused by continual learning. To address these problems, we propose a novel framework called RK2DA, which seamlessly integrates prototype-based data augmentation and relational knowledge distillation. Specifically, RK2DA generates pseudo data by introducing Gaussian noise to the prototype embeddings and utilizes a novel two-phase multi-teacher relational knowledge distillation method to transfer various knowledge from different embedding spaces. Experimental results on the FewRel and TACRED datasets demonstrate that our method outperforms the state-of-the-art baselines.", }

In this paper, we focus on the challenging yet practical problem of Continual Few-shot Relation Extraction (CFRE), which involves extracting relations in the continuous and iterative arrival of new data with only a few labeled examples. The main challenges in CFRE are overfitting due to few-shot learning and catastrophic forgetting caused by continual learning. To address these problems, we propose a novel framework called RK2DA, which seamlessly integrates prototype-based data augmentation and relational knowledge distillation. Specifically, RK2DA generates pseudo data by introducing Gaussian noise to the prototype embeddings and utilizes a novel two-phase multi-teacher relational knowledge distillation method to transfer various knowledge from different embedding spaces. Experimental results on the FewRel and TACRED datasets demonstrate that our method outperforms the state-of-the-art baselines.

Improving Continual Few-shot Relation Extraction through Relational Knowledge Distillation and Prototype Augmentation

lrec-main.767

Poster

https://aclanthology.org/2024.lrec-main.768.bib

https://aclanthology.org/2024.lrec-main.768/

@inproceedings{wu-etal-2024-improving, title = "Improving Copy-oriented Text Generation via {EDU} Copy Mechanism", author = "Wu, Tianxiang and Chen, Han and Qin, Luozheng and Cao, Ziqiang and Ai, Chunhui", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.768", pages = "8768--8780", abstract = "Many text generation tasks are copy-oriented. For instance, nearly 30{\%} content of news summaries is copied. The copy rate is even higher in Grammatical Error Correction (GEC). However, existing generative models generate texts through word-by-word decoding, which may lead to factual inconsistencies and slow inference. While Elementary Discourse Units (EDUs) are outstanding extraction units, EDU-based extractive methods can alleviate the aforementioned problems. As a consequence, we propose EDUCopy, a framework that integrates the behavior of copying EDUs into generative models. The main idea of EDUCopy is to use special index tags to represent the copied EDUs during generation. Specifically, we extract important EDUs from input sequences, finetune generative models to generate sequences with special index tags, and restore the generated special index tags into corresponding text spans. By doing so, EDUCopy reduces the number of generated tokens significantly. To verify the effectiveness of EDUCopy, we conduct experiments on the news summarization datasets CNNDM, NYT and the GEC datasets FCE, WI-LOCNESS. While achieving notable ROUGE and M2 scores, GPT-4 evaluation validates the strength of our models in terms of factual consistency, fluency, and overall performance. Moreover, compared to baseline models, EDUCopy achieves a significant acceleration of 1.65x.", }

Many text generation tasks are copy-oriented. For instance, nearly 30{\%} content of news summaries is copied. The copy rate is even higher in Grammatical Error Correction (GEC). However, existing generative models generate texts through word-by-word decoding, which may lead to factual inconsistencies and slow inference. While Elementary Discourse Units (EDUs) are outstanding extraction units, EDU-based extractive methods can alleviate the aforementioned problems. As a consequence, we propose EDUCopy, a framework that integrates the behavior of copying EDUs into generative models. The main idea of EDUCopy is to use special index tags to represent the copied EDUs during generation. Specifically, we extract important EDUs from input sequences, finetune generative models to generate sequences with special index tags, and restore the generated special index tags into corresponding text spans. By doing so, EDUCopy reduces the number of generated tokens significantly. To verify the effectiveness of EDUCopy, we conduct experiments on the news summarization datasets CNNDM, NYT and the GEC datasets FCE, WI-LOCNESS. While achieving notable ROUGE and M2 scores, GPT-4 evaluation validates the strength of our models in terms of factual consistency, fluency, and overall performance. Moreover, compared to baseline models, EDUCopy achieves a significant acceleration of 1.65x.

Improving Copy-oriented Text Generation via EDU Copy Mechanism

lrec-main.768

Poster

https://aclanthology.org/2024.lrec-main.769.bib

https://aclanthology.org/2024.lrec-main.769/

@inproceedings{li-etal-2024-improving-cross-lingual, title = "Improving Cross-lingual Transfer with Contrastive Negative Learning and Self-training", author = "Li, Guanlin and Zhao, Xuechen and Jafari, Amir and Shao, Wenhao and Farahbakhsh, Reza and Crespi, Noel", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.769", pages = "8781--8791", abstract = "Recent studies improve the cross-lingual transfer learning by better aligning the internal representations within the multilingual model or exploring the information of the target language using self-training. However, the alignment-based methods exhibit intrinsic limitations such as non-transferable linguistic elements, while most of the self-training based methods ignore the useful information hidden in the low-confidence samples. To address this issue, we propose CoNLST (Contrastive Negative Learning and Self-Training) to leverage the information of low-confidence samples. Specifically, we extend the negative learning to the metric space by selecting negative pairs based on the complementary labels and then employ self-training to iteratively train the model to converge on the obtained clean pseudo-labels. We evaluate our approach on the widely-adopted cross-lingual benchmark XNLI. The experiment results show that our method improves upon the baseline models and can serve as a beneficial complement to the alignment-based methods.", }

Recent studies improve the cross-lingual transfer learning by better aligning the internal representations within the multilingual model or exploring the information of the target language using self-training. However, the alignment-based methods exhibit intrinsic limitations such as non-transferable linguistic elements, while most of the self-training based methods ignore the useful information hidden in the low-confidence samples. To address this issue, we propose CoNLST (Contrastive Negative Learning and Self-Training) to leverage the information of low-confidence samples. Specifically, we extend the negative learning to the metric space by selecting negative pairs based on the complementary labels and then employ self-training to iteratively train the model to converge on the obtained clean pseudo-labels. We evaluate our approach on the widely-adopted cross-lingual benchmark XNLI. The experiment results show that our method improves upon the baseline models and can serve as a beneficial complement to the alignment-based methods.

Improving Cross-lingual Transfer with Contrastive Negative Learning and Self-training

lrec-main.769

Poster

https://aclanthology.org/2024.lrec-main.770.bib

https://aclanthology.org/2024.lrec-main.770/

@inproceedings{hu-etal-2024-improving, title = "Improving Factual Consistency in Abstractive Summarization with Sentence Structure Pruning", author = "Hu, Dingxin and Zhang, Xuanyu and Zhang, Xingyue and Li, Yiyang and Chen, Dongsheng and Litvak, Marina and Vanetik, Natalia and Yang, Qing and Xu, Dongliang and Zhou, Yanquan and Li, Lei and Li, Yuze and Zhu, Yingqi", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.770", pages = "8792--8803", abstract = "State-of-the-art abstractive summarization models still suffer from the content contradiction between the summaries and the input text, which is referred to as the factual inconsistency problem. Recently, a large number of works have also been proposed to evaluate factual consistency or improve it by post-editing methods. However, these post-editing methods typically focus on replacing suspicious entities, failing to identify and modify incorrect content hidden in sentence structures. In this paper, we first verify that the correctable errors can be enriched by leveraging sentence structure pruning operation, and then we propose a post-editing method based on that. In the correction process, the pruning operation on possible errors is performed on the syntactic dependency tree with the guidance of multiple factual evaluation metrics. Experimenting on the FRANK dataset shows a great improvement in factual consistency compared with strong baselines and, when combined with them, can achieve even better performance. All the codes and data will be released on paper acceptance.", }

State-of-the-art abstractive summarization models still suffer from the content contradiction between the summaries and the input text, which is referred to as the factual inconsistency problem. Recently, a large number of works have also been proposed to evaluate factual consistency or improve it by post-editing methods. However, these post-editing methods typically focus on replacing suspicious entities, failing to identify and modify incorrect content hidden in sentence structures. In this paper, we first verify that the correctable errors can be enriched by leveraging sentence structure pruning operation, and then we propose a post-editing method based on that. In the correction process, the pruning operation on possible errors is performed on the syntactic dependency tree with the guidance of multiple factual evaluation metrics. Experimenting on the FRANK dataset shows a great improvement in factual consistency compared with strong baselines and, when combined with them, can achieve even better performance. All the codes and data will be released on paper acceptance.

Improving Factual Consistency in Abstractive Summarization with Sentence Structure Pruning

lrec-main.770

Poster

https://aclanthology.org/2024.lrec-main.771.bib

https://aclanthology.org/2024.lrec-main.771/

@inproceedings{li-etal-2024-improving-faithfulness, title = "Improving Faithfulness of Large Language Models in Summarization via Sliding Generation and Self-Consistency", author = "Li, Taiji and Li, Zhi and Zhang, Yin", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.771", pages = "8804--8817", abstract = "Despite large language models (LLMs) have demonstrated impressive performance in various tasks, they are still suffering from the factual inconsistency problem called hallucinations. For instance, LLMs occasionally generate content that diverges from source article, and prefer to extract information that appears at the beginning and end of the context, especially in long document summarization. Inspired by these findings, we propose to improve the faithfulness of LLMs in summarization by impelling them to process the entire article more fairly and faithfully. We present a novel summary generation strategy, namely SliSum, which exploits the ideas of sliding windows and self-consistency. Specifically, SliSum divides the source article into overlapping windows, and utilizes LLM to generate local summaries for the content in the windows. Finally, SliSum aggregates all local summaries using clustering and majority voting algorithm to produce more faithful summary of entire article. Extensive experiments demonstrate that SliSum significantly improves the faithfulness of diverse LLMs including LLaMA-2, Claude-2 and GPT-3.5 in both short and long text summarization, while maintaining their fluency and informativeness and without additional fine-tuning and resources. We further conduct qualitative and quantitative studies to investigate why SliSum works and impacts of hyperparameters in SliSum on performance.", }

Despite large language models (LLMs) have demonstrated impressive performance in various tasks, they are still suffering from the factual inconsistency problem called hallucinations. For instance, LLMs occasionally generate content that diverges from source article, and prefer to extract information that appears at the beginning and end of the context, especially in long document summarization. Inspired by these findings, we propose to improve the faithfulness of LLMs in summarization by impelling them to process the entire article more fairly and faithfully. We present a novel summary generation strategy, namely SliSum, which exploits the ideas of sliding windows and self-consistency. Specifically, SliSum divides the source article into overlapping windows, and utilizes LLM to generate local summaries for the content in the windows. Finally, SliSum aggregates all local summaries using clustering and majority voting algorithm to produce more faithful summary of entire article. Extensive experiments demonstrate that SliSum significantly improves the faithfulness of diverse LLMs including LLaMA-2, Claude-2 and GPT-3.5 in both short and long text summarization, while maintaining their fluency and informativeness and without additional fine-tuning and resources. We further conduct qualitative and quantitative studies to investigate why SliSum works and impacts of hyperparameters in SliSum on performance.

Improving Faithfulness of Large Language Models in Summarization via Sliding Generation and Self-Consistency

lrec-main.771

Poster

2407.21443

https://aclanthology.org/2024.lrec-main.772.bib

https://aclanthology.org/2024.lrec-main.772/

@inproceedings{cao-etal-2024-improving, title = "Improving Grammatical Error Correction by Correction Acceptability Discrimination", author = "Cao, Bin and Jiang, Kai and Pan, Fayu and Bao, Chenlei and Fan, Jing", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.772", pages = "8818--8827", abstract = "Existing Grammatical Error Correction (GEC) methods often overlook the assessment of sentence-level syntax and semantics in the corrected sentence. This oversight results in final corrections that may not be acceptable in the context of the original sentence. In this paper, to improve the performance of Grammatical Error Correction methods, we propose the post-processing task of Correction Acceptability Discrimination (CAD) which aims to remove invalid corrections by comparing the source sentence and its corrected version from the perspective of {``}sentence-level correctness{''}. To solve the CAD task, we propose a pipeline method where the acceptability of each possible correction combination based on the predicted corrections for a source sentence will be judged by a discriminator. Within the discriminator, we design a symmetrical comparison operator to overcome the conflicting results that might be caused by the sentence concatenation order. Experiments show that our method can averagely improve $F_{0.5}$ score by 1.01{\%} over 13 GEC systems in the BEA-2019 test set.", }

Existing Grammatical Error Correction (GEC) methods often overlook the assessment of sentence-level syntax and semantics in the corrected sentence. This oversight results in final corrections that may not be acceptable in the context of the original sentence. In this paper, to improve the performance of Grammatical Error Correction methods, we propose the post-processing task of Correction Acceptability Discrimination (CAD) which aims to remove invalid corrections by comparing the source sentence and its corrected version from the perspective of {``}sentence-level correctness{''}. To solve the CAD task, we propose a pipeline method where the acceptability of each possible correction combination based on the predicted corrections for a source sentence will be judged by a discriminator. Within the discriminator, we design a symmetrical comparison operator to overcome the conflicting results that might be caused by the sentence concatenation order. Experiments show that our method can averagely improve $F_{0.5}$ score by 1.01{\%} over 13 GEC systems in the BEA-2019 test set.

Improving Grammatical Error Correction by Correction Acceptability Discrimination

lrec-main.772

Poster

https://aclanthology.org/2024.lrec-main.773.bib

https://aclanthology.org/2024.lrec-main.773/

@inproceedings{cai-etal-2024-improving, title = "Improving Implicit Discourse Relation Recognition with Semantics Confrontation", author = "Cai, Mingyang and Yang, Zhen and Jian, Ping", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.773", pages = "8828--8839", abstract = "Implicit Discourse Relation Recognition (IDRR), which infers discourse logical relations without explicit connectives, is one of the most challenging tasks in natural language processing (NLP). Recently, pre-trained language models (PLMs) have yielded impressive results across numerous NLP tasks, but their performance still remains unsatisfactory in IDRR. We argue that prior studies have not fully harnessed the potential of PLMs, thereby resulting in a mixture of logical semantics, which determine the logical relations between discourse arguments, and general semantics, which encapsulate the non-logical contextual aspects (detailed in Sec.1). Such a mixture would inevitably compromise the logic reasoning ability of PLMs. Therefore, we propose a novel method that trains the PLMs through two semantics enhancers to implicitly differentiate logical and general semantics, ultimately achieving logical semantics enhancement. Due to the characteristic of PLM in word representation learning, these two semantics enhancers will inherently confront with each other, facilitating an augmentation of logical semantics by disentangling them from general semantics. The experimental results on PDTB 2.0 dataset show that the confrontation approach exceeds our baseline by 3.81{\%} F1 score, and the effectiveness of the semantics confrontation method is validated by comprehensive ablation experiments.", }

Implicit Discourse Relation Recognition (IDRR), which infers discourse logical relations without explicit connectives, is one of the most challenging tasks in natural language processing (NLP). Recently, pre-trained language models (PLMs) have yielded impressive results across numerous NLP tasks, but their performance still remains unsatisfactory in IDRR. We argue that prior studies have not fully harnessed the potential of PLMs, thereby resulting in a mixture of logical semantics, which determine the logical relations between discourse arguments, and general semantics, which encapsulate the non-logical contextual aspects (detailed in Sec.1). Such a mixture would inevitably compromise the logic reasoning ability of PLMs. Therefore, we propose a novel method that trains the PLMs through two semantics enhancers to implicitly differentiate logical and general semantics, ultimately achieving logical semantics enhancement. Due to the characteristic of PLM in word representation learning, these two semantics enhancers will inherently confront with each other, facilitating an augmentation of logical semantics by disentangling them from general semantics. The experimental results on PDTB 2.0 dataset show that the confrontation approach exceeds our baseline by 3.81{\%} F1 score, and the effectiveness of the semantics confrontation method is validated by comprehensive ablation experiments.

Improving Implicit Discourse Relation Recognition with Semantics Confrontation

lrec-main.773

Poster

https://aclanthology.org/2024.lrec-main.774.bib

https://aclanthology.org/2024.lrec-main.774/

@inproceedings{feng-etal-2024-improving, title = "Improving Language Model Reasoning with Self-motivated Learning", author = "Feng, Yunlong and Xu, Yang and Qin, Libo and Wang, Yasheng and Che, Wanxiang", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.774", pages = "8840--8852", abstract = "Large-scale high-quality training data is important for improving the performance of models. After trained with data that has rationales (reasoning steps), models gain reasoning capability. However, the dataset with high-quality rationales is relatively scarce due to the high annotation cost. To address this issue, we propose \textit{Self-motivated Learning} framework. The framework motivates the model itself to automatically generate rationales on existing datasets. Based on the inherent rank from correctness across multiple rationales, the model learns to generate better rationales, leading to higher reasoning capability. Specifically, we train a reward model with the rank to evaluate the quality of rationales, and improve the performance of reasoning through reinforcement learning. Experiment results of Llama2 7B on multiple reasoning datasets show that our method significantly improves the reasoning ability of models, even outperforming InstructGPT in some datasets.", }

Large-scale high-quality training data is important for improving the performance of models. After trained with data that has rationales (reasoning steps), models gain reasoning capability. However, the dataset with high-quality rationales is relatively scarce due to the high annotation cost. To address this issue, we propose \textit{Self-motivated Learning} framework. The framework motivates the model itself to automatically generate rationales on existing datasets. Based on the inherent rank from correctness across multiple rationales, the model learns to generate better rationales, leading to higher reasoning capability. Specifically, we train a reward model with the rank to evaluate the quality of rationales, and improve the performance of reasoning through reinforcement learning. Experiment results of Llama2 7B on multiple reasoning datasets show that our method significantly improves the reasoning ability of models, even outperforming InstructGPT in some datasets.

Improving Language Model Reasoning with Self-motivated Learning

lrec-main.774

Poster

2404.07017

https://aclanthology.org/2024.lrec-main.775.bib

https://aclanthology.org/2024.lrec-main.775/

@inproceedings{kang-shin-2024-improving, title = "Improving Low-Resource Keyphrase Generation through Unsupervised Title Phrase Generation", author = "Kang, Byungha and Shin, Youhyun", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.775", pages = "8853--8865", abstract = "This paper introduces a novel approach called title phrase generation (TPG) for unsupervised keyphrase generation (UKG), leveraging a pseudo label generated from a document title. Previous UKG method extracts all phrases from a corpus to build a phrase bank, then draws candidate absent keyphrases related to a document from the phrase bank to generate a pseudo label. However, we observed that when separating the document title from the document body, a significant number of phrases absent from the document body are included in the title. Based on this observation, we propose an effective method for generating pseudo labels using phrases mined from the document title. We initially train BART using these pseudo labels (TPG) and then perform supervised fine-tuning on a small amount of human-annotated data, which we term low-resource fine-tuning (LRFT). Experimental results on five benchmark datasets demonstrate that our method outperforms existing low-resource keyphrase generation approaches even with fewer labeled data, showing strength in generating absent keyphrases. Moreover, our model trained solely with TPG, without any labeled data, surpasses previous UKG method, highlighting the effectiveness of utilizing titles over a phrase bank. The code is available at https://github.com/kangnlp/low-resource-kpgen-through-TPG.", }

This paper introduces a novel approach called title phrase generation (TPG) for unsupervised keyphrase generation (UKG), leveraging a pseudo label generated from a document title. Previous UKG method extracts all phrases from a corpus to build a phrase bank, then draws candidate absent keyphrases related to a document from the phrase bank to generate a pseudo label. However, we observed that when separating the document title from the document body, a significant number of phrases absent from the document body are included in the title. Based on this observation, we propose an effective method for generating pseudo labels using phrases mined from the document title. We initially train BART using these pseudo labels (TPG) and then perform supervised fine-tuning on a small amount of human-annotated data, which we term low-resource fine-tuning (LRFT). Experimental results on five benchmark datasets demonstrate that our method outperforms existing low-resource keyphrase generation approaches even with fewer labeled data, showing strength in generating absent keyphrases. Moreover, our model trained solely with TPG, without any labeled data, surpasses previous UKG method, highlighting the effectiveness of utilizing titles over a phrase bank. The code is available at https://github.com/kangnlp/low-resource-kpgen-through-TPG.

Improving Low-Resource Keyphrase Generation through Unsupervised Title Phrase Generation

lrec-main.775

Poster

https://aclanthology.org/2024.lrec-main.776.bib

https://aclanthology.org/2024.lrec-main.776/

@inproceedings{bai-bai-2024-improving, title = "Improving Multi-view Document Clustering: Leveraging Multi-structure Processor and Hybrid Ensemble Clustering Module", author = "Bai, Ruina and Bai, Qi", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.776", pages = "8866--8876", abstract = "We introduce a multi-view document clustering model called DMsECN (Deep Multi-structure Ensemble Clustering Network), comprising a multi-structure processor and a hybrid ensemble clustering module. Unlike existing models, DMsECN distinguishes itself by creating a consensus structure from multiple clustering structures. The multi-structure processor comprises two stages, each contributing to the extraction of clustering structures that preserve both consistency and complementarity across multiple views. Representation learning extracts both view and view-fused representations from multi-views through the use of contrastive learning. Subsequently, multi-structure learning employs distinct view clustering guidance to generate the corresponding clustering structures. The hybrid ensemble clustering module merges two ensemble methods to amalgamate multiple structures, producing a consensus structure that guarantees both the separability and compactness of clusters within the clustering results. The attention-based ensemble primarily concentrates on learning the contribution weights of diverse clustering structures, while the similarity-based ensemble employs cluster assignment similarity and cluster classification dissimilarity to guide the refinement of the consensus structure. Experimental results demonstrate that DMsECN outperforms other models, achieving new state-of-the-art results on four multi-view document clustering datasets.", }

We introduce a multi-view document clustering model called DMsECN (Deep Multi-structure Ensemble Clustering Network), comprising a multi-structure processor and a hybrid ensemble clustering module. Unlike existing models, DMsECN distinguishes itself by creating a consensus structure from multiple clustering structures. The multi-structure processor comprises two stages, each contributing to the extraction of clustering structures that preserve both consistency and complementarity across multiple views. Representation learning extracts both view and view-fused representations from multi-views through the use of contrastive learning. Subsequently, multi-structure learning employs distinct view clustering guidance to generate the corresponding clustering structures. The hybrid ensemble clustering module merges two ensemble methods to amalgamate multiple structures, producing a consensus structure that guarantees both the separability and compactness of clusters within the clustering results. The attention-based ensemble primarily concentrates on learning the contribution weights of diverse clustering structures, while the similarity-based ensemble employs cluster assignment similarity and cluster classification dissimilarity to guide the refinement of the consensus structure. Experimental results demonstrate that DMsECN outperforms other models, achieving new state-of-the-art results on four multi-view document clustering datasets.

Improving Multi-view Document Clustering: Leveraging Multi-structure Processor and Hybrid Ensemble Clustering Module

lrec-main.776

Poster

https://aclanthology.org/2024.lrec-main.777.bib

https://aclanthology.org/2024.lrec-main.777/

@inproceedings{zhang-etal-2024-improving-personalized, title = "Improving Personalized Sentiment Representation with Knowledge-enhanced and Parameter-efficient Layer Normalization", author = "Zhang, You and Wang, Jin and Yu, Liang-Chih and Xu, Dan and Zhang, Xuejie", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.777", pages = "8877--8889", abstract = "Existing studies on personalized sentiment classification consider a document review as an overall text unit and incorporate backgrounds (i.e., user and product information) to learn sentiment representation. However, it is difficult when these methods meet the current pretrained language models (PLMs) owing to quadratic costs that increase with text length and heterogeneous mixes of randomly initialized background information and textual information initialized from well-pretrained checkpoints during information incorporation. To address these problems, we propose a knowledge-enhanced and parameter-efficient layer normalization (E2LN) for efficient and effective review modeling via leveraging LN in transformer structures. Initially, a knowledge base is introduced that stores well-pretrained checkpoints, structured text information, and background information. Based on such a knowledge base, the ability of LN can be magnified as being a crucial component of transformer structure and then improve the performance of PLMs in downstream tasks. Moreover, the proposed E2LN can make PLMs capable of modeling long document reviews and incorporating background information with parameter-efficient fine-tuning and knowledge injecting. Extensive experimental results were obtained for three document-level sentiment classification benchmark datasets. By comparing the results, the effectiveness and efficiency of the proposed model was demonstrated. Code and Data are released at https://github.com/yoyo-yun/E2LN.", }

Existing studies on personalized sentiment classification consider a document review as an overall text unit and incorporate backgrounds (i.e., user and product information) to learn sentiment representation. However, it is difficult when these methods meet the current pretrained language models (PLMs) owing to quadratic costs that increase with text length and heterogeneous mixes of randomly initialized background information and textual information initialized from well-pretrained checkpoints during information incorporation. To address these problems, we propose a knowledge-enhanced and parameter-efficient layer normalization (E2LN) for efficient and effective review modeling via leveraging LN in transformer structures. Initially, a knowledge base is introduced that stores well-pretrained checkpoints, structured text information, and background information. Based on such a knowledge base, the ability of LN can be magnified as being a crucial component of transformer structure and then improve the performance of PLMs in downstream tasks. Moreover, the proposed E2LN can make PLMs capable of modeling long document reviews and incorporating background information with parameter-efficient fine-tuning and knowledge injecting. Extensive experimental results were obtained for three document-level sentiment classification benchmark datasets. By comparing the results, the effectiveness and efficiency of the proposed model was demonstrated. Code and Data are released at https://github.com/yoyo-yun/E2LN.

Improving Personalized Sentiment Representation with Knowledge-enhanced and Parameter-efficient Layer Normalization

lrec-main.777

Poster

https://aclanthology.org/2024.lrec-main.778.bib

https://aclanthology.org/2024.lrec-main.778/

@inproceedings{ding-etal-2024-improving, title = "Improving Recall of Large Language Models: A Model Collaboration Approach for Relational Triple Extraction", author = "Ding, Zepeng and Huang, Wenhao and Liang, Jiaqing and Xiao, Yanghua and Yang, Deqing", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.778", pages = "8890--8901", abstract = "Relation triple extraction, which outputs a set of triples from long sentences, plays a vital role in knowledge acquisition. Large language models can accurately extract triples from simple sentences through few-shot learning or fine-tuning when given appropriate instructions. However, they often miss out when extracting from complex sentences. In this paper, we design an evaluation-filtering framework that integrates large language models with small models for relational triple extraction tasks. The framework includes an evaluation model that can extract related entity pairs with high precision. We propose a simple labeling principle and a deep neural network to build the model, embedding the outputs as prompts into the extraction process of the large model. We conduct extensive experiments to demonstrate that the proposed method can assist large language models in obtaining more accurate extraction results, especially from complex sentences containing multiple relational triples. Our evaluation model can also be embedded into traditional extraction models to enhance their extraction precision from complex sentences.", }

Relation triple extraction, which outputs a set of triples from long sentences, plays a vital role in knowledge acquisition. Large language models can accurately extract triples from simple sentences through few-shot learning or fine-tuning when given appropriate instructions. However, they often miss out when extracting from complex sentences. In this paper, we design an evaluation-filtering framework that integrates large language models with small models for relational triple extraction tasks. The framework includes an evaluation model that can extract related entity pairs with high precision. We propose a simple labeling principle and a deep neural network to build the model, embedding the outputs as prompts into the extraction process of the large model. We conduct extensive experiments to demonstrate that the proposed method can assist large language models in obtaining more accurate extraction results, especially from complex sentences containing multiple relational triples. Our evaluation model can also be embedded into traditional extraction models to enhance their extraction precision from complex sentences.

Improving Recall of Large Language Models: A Model Collaboration Approach for Relational Triple Extraction

lrec-main.778

Poster

2404.09593

https://huggingface.co/papers/2404.09593

https://aclanthology.org/2024.lrec-main.779.bib

https://aclanthology.org/2024.lrec-main.779/

@inproceedings{zheng-etal-2024-improving, title = "Improving Robustness of {GNN}-based Anomaly Detection by Graph Adversarial Training", author = "Zheng, Xiangping and Wu, Bo and Zhang, Alex X. and Li, Wei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.779", pages = "8902--8912", abstract = "Graph neural networks (GNNs) play a fundamental role in anomaly detection, excelling at the identification of node anomalies by aggregating information from neighboring nodes. Nonetheless, they exhibit vulnerability to attacks, with even minor alterations in the graph structure or node attributes resulting in substantial performance degradation. To address this critical challenge, we introduce an innovative mechanism for graph adversarial training, meticulously designed to bolster GNN-based anomaly detection systems against potential poisoning attacks. This novel approach follows a two-step framework. (1) In the initial phase, we employ a Multiple-Objective Generative Adversarial Attack (MO-GAA), which focuses on generating feature modifications and inducing structural disruptions within the graph. Its primary objective is to mimic the adversarial behavior of potential attackers on the anomaly detection graph, with the explicit intention of confounding the anomaly detector. (2) In the subsequent stage, we introduce Purification-Based Adversarial Attack Defense (PB-AAD), a method specifically designed to rectify any contamination and restore the integrity of the graph. The central aim of PB-AAD is to counteract the destructive actions carried out by potential attackers. Our empirical findings, derived from extensive experiments conducted on four real-world anomaly detection datasets, serve to demonstrate how MO-GAA systematically disrupts the graph, compromising the effectiveness of GNN-based detectors, while PB-AAD effectively mitigates these adversarial actions, thereby enhancing the overall robustness of GNN-based anomaly detectors.", }

Graph neural networks (GNNs) play a fundamental role in anomaly detection, excelling at the identification of node anomalies by aggregating information from neighboring nodes. Nonetheless, they exhibit vulnerability to attacks, with even minor alterations in the graph structure or node attributes resulting in substantial performance degradation. To address this critical challenge, we introduce an innovative mechanism for graph adversarial training, meticulously designed to bolster GNN-based anomaly detection systems against potential poisoning attacks. This novel approach follows a two-step framework. (1) In the initial phase, we employ a Multiple-Objective Generative Adversarial Attack (MO-GAA), which focuses on generating feature modifications and inducing structural disruptions within the graph. Its primary objective is to mimic the adversarial behavior of potential attackers on the anomaly detection graph, with the explicit intention of confounding the anomaly detector. (2) In the subsequent stage, we introduce Purification-Based Adversarial Attack Defense (PB-AAD), a method specifically designed to rectify any contamination and restore the integrity of the graph. The central aim of PB-AAD is to counteract the destructive actions carried out by potential attackers. Our empirical findings, derived from extensive experiments conducted on four real-world anomaly detection datasets, serve to demonstrate how MO-GAA systematically disrupts the graph, compromising the effectiveness of GNN-based detectors, while PB-AAD effectively mitigates these adversarial actions, thereby enhancing the overall robustness of GNN-based anomaly detectors.

Improving Robustness of GNN-based Anomaly Detection by Graph Adversarial Training

lrec-main.779

Poster

https://aclanthology.org/2024.lrec-main.780.bib

https://aclanthology.org/2024.lrec-main.780/

@inproceedings{guan-etal-2024-improving, title = "Improving Role-Oriented Dialogue Summarization with Interaction-Aware Contrastive Learning", author = "Guan, Weihong and Feng, Shi and Wang, Daling and Huang, Faliang and Zhang, Yifei and Cui, Yuan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.780", pages = "8913--8924", abstract = "Role-oriented dialogue summarization aims at generating summaries for different roles in dialogue, e.g., user and agent. Interaction between different roles is vital for the task. Existing methods could not fully capture interaction patterns between roles when encoding dialogue, thus are prone to ignore the interaction-related key information. In this paper, we propose a contrastive learning based interaction-aware model for the role-oriented dialogue summarization namely CIAM. An interaction-aware contrastive objective is constructed to guide the encoded dialogue representation to learn role-level interaction. The representation is then used by the decoder to generate role-oriented summaries. The contrastive objective is trained jointly with the primary dialogue summarization task. Additionally, we innovatively utilize different decoder start tokens to control what kind of summary to generate, thus could generate different role-oriented summaries with a unified model. Experimental results show that our method achieves new state-of-the-art results on two public datasets. Extensive analyses further demonstrate that our method excels at capturing interaction information between different roles and producing informative summaries.", }

Role-oriented dialogue summarization aims at generating summaries for different roles in dialogue, e.g., user and agent. Interaction between different roles is vital for the task. Existing methods could not fully capture interaction patterns between roles when encoding dialogue, thus are prone to ignore the interaction-related key information. In this paper, we propose a contrastive learning based interaction-aware model for the role-oriented dialogue summarization namely CIAM. An interaction-aware contrastive objective is constructed to guide the encoded dialogue representation to learn role-level interaction. The representation is then used by the decoder to generate role-oriented summaries. The contrastive objective is trained jointly with the primary dialogue summarization task. Additionally, we innovatively utilize different decoder start tokens to control what kind of summary to generate, thus could generate different role-oriented summaries with a unified model. Experimental results show that our method achieves new state-of-the-art results on two public datasets. Extensive analyses further demonstrate that our method excels at capturing interaction information between different roles and producing informative summaries.

Improving Role-Oriented Dialogue Summarization with Interaction-Aware Contrastive Learning

lrec-main.780

Poster

https://aclanthology.org/2024.lrec-main.781.bib

https://aclanthology.org/2024.lrec-main.781/

@inproceedings{jamelot-etal-2024-improving, title = "Improving Text Readability through Segmentation into Rheses", author = "Jamelot, Antoine and Quiniou, Solen and Hamon, Sophie", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.781", pages = "8925--8930", abstract = "Enhancing text readability is crucial for readers with challenges like dyslexia. This paper delves into the segmentation of sentences into rheses, i.e. rhythmic and semantic units. Their aim is to clarify sentence structures for improved comprehension, through a harmonious balance between syntactic accuracy, the natural rhythm of reading aloud, and the delineation of meaningful units. This study relates and compares our various attempts to improve a pre-existing rhesis segmentation tool, which is based on the selection of candidate segmentations. We also release TeRheSe (Texts with Rhesis Segmentation), a bilingual dataset, segmented into rheses, comprising 12 books from classic literature in French and English. We evaluated our approaches on this dataset, showing the efficiency of a novel approach based on token classification, reaching a F1-score of 90.0{\%} in English (previously 85.3{\%}) and 91.3{\%} in French (previously 88.0{\%}). We also study the potential of leveraging prosodic elements, though its definitive impact remains inconclusive.", }

Enhancing text readability is crucial for readers with challenges like dyslexia. This paper delves into the segmentation of sentences into rheses, i.e. rhythmic and semantic units. Their aim is to clarify sentence structures for improved comprehension, through a harmonious balance between syntactic accuracy, the natural rhythm of reading aloud, and the delineation of meaningful units. This study relates and compares our various attempts to improve a pre-existing rhesis segmentation tool, which is based on the selection of candidate segmentations. We also release TeRheSe (Texts with Rhesis Segmentation), a bilingual dataset, segmented into rheses, comprising 12 books from classic literature in French and English. We evaluated our approaches on this dataset, showing the efficiency of a novel approach based on token classification, reaching a F1-score of 90.0{\%} in English (previously 85.3{\%}) and 91.3{\%} in French (previously 88.0{\%}). We also study the potential of leveraging prosodic elements, though its definitive impact remains inconclusive.

Improving Text Readability through Segmentation into Rheses

lrec-main.781

Poster

https://aclanthology.org/2024.lrec-main.782.bib

https://aclanthology.org/2024.lrec-main.782/

@inproceedings{zhao-etal-2024-improving, title = "Improving the Robustness of Large Language Models via Consistency Alignment", author = "Zhao, Yukun and Yan, Lingyong and Sun, Weiwei and Xing, Guoliang and Wang, Shuaiqiang and Meng, Chong and Cheng, Zhicong and Ren, Zhaochun and Yin, Dawei", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.782", pages = "8931--8941", abstract = "Large language models (LLMs) have shown tremendous success in following user instructions and generating helpful responses. Nevertheless, their robustness is still far from optimal, as they may generate significantly inconsistent responses due to minor changes in the verbalized instructions. Recent literature has explored this inconsistency issue, highlighting the importance of continued improvement in the robustness of response generation. However, systematic analysis and solutions are still lacking. In this paper, we quantitatively define the inconsistency problem and propose a two-stage training framework consisting of instruction-augmented supervised fine-tuning and consistency alignment training. The first stage helps a model generalize on following instructions via similar instruction augmentations. In the second stage, we improve the diversity and help the model understand which responses are more aligned with human expectations by differentiating subtle differences in similar responses. The training process is accomplished by self-rewards inferred from the trained model at the first stage without referring to external human preference resources. We conduct extensive experiments on recent publicly available LLMs on instruction-following tasks and demonstrate the effectiveness of our training framework.", }

Large language models (LLMs) have shown tremendous success in following user instructions and generating helpful responses. Nevertheless, their robustness is still far from optimal, as they may generate significantly inconsistent responses due to minor changes in the verbalized instructions. Recent literature has explored this inconsistency issue, highlighting the importance of continued improvement in the robustness of response generation. However, systematic analysis and solutions are still lacking. In this paper, we quantitatively define the inconsistency problem and propose a two-stage training framework consisting of instruction-augmented supervised fine-tuning and consistency alignment training. The first stage helps a model generalize on following instructions via similar instruction augmentations. In the second stage, we improve the diversity and help the model understand which responses are more aligned with human expectations by differentiating subtle differences in similar responses. The training process is accomplished by self-rewards inferred from the trained model at the first stage without referring to external human preference resources. We conduct extensive experiments on recent publicly available LLMs on instruction-following tasks and demonstrate the effectiveness of our training framework.

Improving the Robustness of Large Language Models via Consistency Alignment

lrec-main.782

Poster

2403.14221

https://huggingface.co/papers/2403.14221

https://aclanthology.org/2024.lrec-main.783.bib

https://aclanthology.org/2024.lrec-main.783/

@inproceedings{lu-zhang-2024-improving, title = "Improving Unsupervised Neural Machine Translation via Training Data Self-Correction", author = "Lu, Jinliang and Zhang, Jiajun", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.783", pages = "8942--8954", abstract = "Unsupervised neural machine translation (UNMT) models are trained with pseudo-parallel sentences constructed by on-the-fly back-translation using monolingual corpora. However, the quality of pseudo-parallel sentences cannot be guaranteed, which hinders the final performance of UNMT. This paper demonstrates that although UNMT usually generates mistakes during pseudo-parallel data construction, some of them can be corrected by the token-level translations that exist in the embedding table. Therefore, we propose a self-correction method to automatically improve the quality of pseudo-parallel sentences during training, thereby enhancing translation performance. Specifically, for a pseudo sentence pair, our self-correction method first estimates the alignment relations between tokens by treating and solving it as an optimal transport problem. Then, we measure the translation reliability for each token and detect the mis-translated ones. Finally, the mis-translated tokens are corrected with real-time computed token-by-token translations based on the embedding table, yielding a better training example. Considering that the modified examples are semantically equivalent to the original ones when UNMT converges, we introduce second-phase training to strengthen the output consistency between them, further improving the generalization capability and translation performance. Empirical results on widely used UNMT datasets demonstrate the effectiveness of our method and it significantly outperforms several strong baselines.", }

Unsupervised neural machine translation (UNMT) models are trained with pseudo-parallel sentences constructed by on-the-fly back-translation using monolingual corpora. However, the quality of pseudo-parallel sentences cannot be guaranteed, which hinders the final performance of UNMT. This paper demonstrates that although UNMT usually generates mistakes during pseudo-parallel data construction, some of them can be corrected by the token-level translations that exist in the embedding table. Therefore, we propose a self-correction method to automatically improve the quality of pseudo-parallel sentences during training, thereby enhancing translation performance. Specifically, for a pseudo sentence pair, our self-correction method first estimates the alignment relations between tokens by treating and solving it as an optimal transport problem. Then, we measure the translation reliability for each token and detect the mis-translated ones. Finally, the mis-translated tokens are corrected with real-time computed token-by-token translations based on the embedding table, yielding a better training example. Considering that the modified examples are semantically equivalent to the original ones when UNMT converges, we introduce second-phase training to strengthen the output consistency between them, further improving the generalization capability and translation performance. Empirical results on widely used UNMT datasets demonstrate the effectiveness of our method and it significantly outperforms several strong baselines.

Improving Unsupervised Neural Machine Translation via Training Data Self-Correction

lrec-main.783

Poster

https://aclanthology.org/2024.lrec-main.784.bib

https://aclanthology.org/2024.lrec-main.784/

@inproceedings{vo-etal-2024-improving, title = "Improving {V}ietnamese-{E}nglish Medical Machine Translation", author = "Vo, Nhu and Nguyen, Dat Quoc and Le, Dung D. and Piccardi, Massimo and Buntine, Wray", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.784", pages = "8955--8962", abstract = "Machine translation for Vietnamese-English in the medical domain is still an under-explored research area. In this paper, we introduce MedEV{---}a high-quality Vietnamese-English parallel dataset constructed specifically for the medical domain, comprising approximately 360K sentence pairs. We conduct extensive experiments comparing Google Translate, ChatGPT (gpt-3.5-turbo), state-of-the-art Vietnamese-English neural machine translation models and pre-trained bilingual/multilingual sequence-to-sequence models on our new MedEV dataset. Experimental results show that the best performance is achieved by fine-tuning {``}vinai-translate{''} for each translation direction. We publicly release our dataset to promote further research.", }

Machine translation for Vietnamese-English in the medical domain is still an under-explored research area. In this paper, we introduce MedEV{---}a high-quality Vietnamese-English parallel dataset constructed specifically for the medical domain, comprising approximately 360K sentence pairs. We conduct extensive experiments comparing Google Translate, ChatGPT (gpt-3.5-turbo), state-of-the-art Vietnamese-English neural machine translation models and pre-trained bilingual/multilingual sequence-to-sequence models on our new MedEV dataset. Experimental results show that the best performance is achieved by fine-tuning {``}vinai-translate{''} for each translation direction. We publicly release our dataset to promote further research.

Improving Vietnamese-English Medical Machine Translation

lrec-main.784

Poster

2403.19161

https://huggingface.co/papers/2403.19161

https://aclanthology.org/2024.lrec-main.785.bib

https://aclanthology.org/2024.lrec-main.785/

@inproceedings{doukhan-etal-2024-inagvad, title = "{I}na{GVAD} : A Challenging {F}rench {TV} and Radio Corpus Annotated for Speech Activity Detection and Speaker Gender Segmentation", author = "Doukhan, David and Maertens, Christine and Le Personnic, William and Speroni, Ludovic and Dehak, Reda", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.785", pages = "8963--8974", abstract = "InaGVAD is an audio corpus collected from 10 French radio and 18 TV channels categorized into 4 groups: generalist radio, music radio, news TV, and generalist TV. It contains 277 1-minute-long annotated recordings aimed at representing the acoustic diversity of French audiovisual programs and was primarily designed to build systems able to monitor men{'}s and women{'}s speaking time in media. inaGVAD is provided with Voice Activity Detection (VAD) and Speaker Gender Segmentation (SGS) annotations extended with overlap, speaker traits (gender, age, voice quality), and 10 non-speech event categories. Annotation distributions are detailed for each channel category. This dataset is partitioned into a 1h development and a 3h37 test subset, allowing fair and reproducible system evaluation. A benchmark of 6 freely available VAD software is presented, showing diverse abilities based on channel and non-speech event categories. Two existing SGS systems are evaluated on the corpus and compared against a baseline X-vector transfer learning strategy, trained on the development subset. Results demonstrate that our proposal, trained on a single - but diverse - hour of data, achieved competitive SGS results. The entire inaGVAD package; including corpus, annotations, evaluation scripts, and baseline training code; is made freely accessible, fostering future advancement in the domain.", }

InaGVAD is an audio corpus collected from 10 French radio and 18 TV channels categorized into 4 groups: generalist radio, music radio, news TV, and generalist TV. It contains 277 1-minute-long annotated recordings aimed at representing the acoustic diversity of French audiovisual programs and was primarily designed to build systems able to monitor men{'}s and women{'}s speaking time in media. inaGVAD is provided with Voice Activity Detection (VAD) and Speaker Gender Segmentation (SGS) annotations extended with overlap, speaker traits (gender, age, voice quality), and 10 non-speech event categories. Annotation distributions are detailed for each channel category. This dataset is partitioned into a 1h development and a 3h37 test subset, allowing fair and reproducible system evaluation. A benchmark of 6 freely available VAD software is presented, showing diverse abilities based on channel and non-speech event categories. Two existing SGS systems are evaluated on the corpus and compared against a baseline X-vector transfer learning strategy, trained on the development subset. Results demonstrate that our proposal, trained on a single - but diverse - hour of data, achieved competitive SGS results. The entire inaGVAD package; including corpus, annotations, evaluation scripts, and baseline training code; is made freely accessible, fostering future advancement in the domain.

InaGVAD : A Challenging French TV and Radio Corpus Annotated for Speech Activity Detection and Speaker Gender Segmentation

lrec-main.785

Poster

2406.04429

https://aclanthology.org/2024.lrec-main.786.bib

https://aclanthology.org/2024.lrec-main.786/

@inproceedings{wang-etal-2024-context, title = "In-Context Example Retrieval from Multi-Perspectives for Few-Shot Aspect-Based Sentiment Analysis", author = "Wang, Qianlong and Xu, Hongling and Ding, Keyang and Liang, Bin and Xu, Ruifeng", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.786", pages = "8975--8985", abstract = "In this paper, we focus on few-shot aspect-based sentiment analysis (ABSA) and try to solve it with in-context learning (ICL) paradigm. However, the effectiveness of ICL is highly affected by retrieved in-context examples. Previous works generally leverage the semantic similarity between the candidate examples and test input to retrieve examples. However, they may yield sub-optimal results for this task. This is because considering only the overall semantic perspective may leave some useful examples, which have syntactic structural relevance to the test input or share identical sentiments and similar aspects to one unretrievable. To address this shortcoming, we advocate retrieving in-context examples for few-shot ABSA by simultaneously considering three perspectives, overall semantics, syntactic structure relevance, and aspect-sentiment semantics. To achieve this, we construct positive and negative pairs from these three perspectives and train the demonstration retriever using contrastive learning. Experimental results on four ABSA datasets show that our retrieval framework can significantly outperform baselines across the board. Moreover, to understand factors influencing ICL performance on few-shot ABSA, we conduct extensive analysis in various scenarios, which can inspire and advance future research.", }

In this paper, we focus on few-shot aspect-based sentiment analysis (ABSA) and try to solve it with in-context learning (ICL) paradigm. However, the effectiveness of ICL is highly affected by retrieved in-context examples. Previous works generally leverage the semantic similarity between the candidate examples and test input to retrieve examples. However, they may yield sub-optimal results for this task. This is because considering only the overall semantic perspective may leave some useful examples, which have syntactic structural relevance to the test input or share identical sentiments and similar aspects to one unretrievable. To address this shortcoming, we advocate retrieving in-context examples for few-shot ABSA by simultaneously considering three perspectives, overall semantics, syntactic structure relevance, and aspect-sentiment semantics. To achieve this, we construct positive and negative pairs from these three perspectives and train the demonstration retriever using contrastive learning. Experimental results on four ABSA datasets show that our retrieval framework can significantly outperform baselines across the board. Moreover, to understand factors influencing ICL performance on few-shot ABSA, we conduct extensive analysis in various scenarios, which can inspire and advance future research.

In-Context Example Retrieval from Multi-Perspectives for Few-Shot Aspect-Based Sentiment Analysis

lrec-main.786

Poster

https://aclanthology.org/2024.lrec-main.787.bib

https://aclanthology.org/2024.lrec-main.787/

@inproceedings{zheng-etal-2024-incorporating, title = "Incorporating Lexical and Syntactic Knowledge for Unsupervised Cross-Lingual Transfer", author = "Zheng, Jianyu and Fan, Fengfei and Li, Jianquan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.787", pages = "8986--8997", abstract = "Unsupervised cross-lingual transfer involves transferring knowledge between languages without explicit supervision. Although numerous studies have been conducted to improve performance in such tasks by focusing on cross-lingual knowledge, particularly lexical and syntactic knowledge, current approaches are limited as they only incorporate syntactic or lexical information. Since each type of information offers unique advantages and no previous attempts have combined both, we attempt to explore the potential of this approach. In this paper, we present a novel framework called {``}Lexicon-Syntax Enhanced Multilingual BERT{''} that combines both lexical and syntactic knowledge. Specifically, we use Multilingual BERT (mBERT) as the base model and employ two techniques to enhance its learning capabilities. The code-switching technique is used to implicitly teach the model lexical alignment information, while a syntactic-based graph attention network is designed to help the model encode syntactic structure. To integrate both types of knowledge, we input code-switched sequences into both the syntactic module and the mBERT base model simultaneously. Our extensive experimental results demonstrate this framework can consistently outperform all baselines of zero-shot cross-lingual transfer, with the gains of 1.0 3.7 points on text classification, named entity recognition (ner), and semantic parsing tasks.", }

Unsupervised cross-lingual transfer involves transferring knowledge between languages without explicit supervision. Although numerous studies have been conducted to improve performance in such tasks by focusing on cross-lingual knowledge, particularly lexical and syntactic knowledge, current approaches are limited as they only incorporate syntactic or lexical information. Since each type of information offers unique advantages and no previous attempts have combined both, we attempt to explore the potential of this approach. In this paper, we present a novel framework called {``}Lexicon-Syntax Enhanced Multilingual BERT{''} that combines both lexical and syntactic knowledge. Specifically, we use Multilingual BERT (mBERT) as the base model and employ two techniques to enhance its learning capabilities. The code-switching technique is used to implicitly teach the model lexical alignment information, while a syntactic-based graph attention network is designed to help the model encode syntactic structure. To integrate both types of knowledge, we input code-switched sequences into both the syntactic module and the mBERT base model simultaneously. Our extensive experimental results demonstrate this framework can consistently outperform all baselines of zero-shot cross-lingual transfer, with the gains of 1.0 3.7 points on text classification, named entity recognition (ner), and semantic parsing tasks.

Incorporating Lexical and Syntactic Knowledge for Unsupervised Cross-Lingual Transfer

lrec-main.787

Poster

2404.16627

https://aclanthology.org/2024.lrec-main.788.bib

https://aclanthology.org/2024.lrec-main.788/

@inproceedings{pinney-etal-2024-incorporating, title = "Incorporating Word-level Phonemic Decoding into Readability Assessment", author = "Pinney, Christine and Kennington, Casey and Pera, Maria Soledad and Landau Wright, Katherine and Fails, Jerry Alan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.788", pages = "8998--9009", abstract = "Current approaches in automatic readability assessment have found success with the use of large language models and transformer architectures. These techniques lead to accuracy improvement, but they do not offer the interpretability that is uniquely required by the audience most often employing readability assessment tools: teachers and educators. Recent work that employs more traditional machine learning methods has highlighted the linguistic importance of considering semantic and syntactic characteristics of text in readability assessment by utilizing handcrafted feature sets. Research in Education suggests that, in addition to semantics and syntax, phonetic and orthographic instruction are necessary for children to progress through the stages of reading and spelling development; children must first learn to decode the letters and symbols on a page to recognize words and phonemes and their connection to speech sounds. Here, we incorporate this word-level phonemic decoding process into readability assessment by crafting a phonetically-based feature set for grade-level classification for English. Our resulting feature set shows comparable performance to much larger, semantically- and syntactically-based feature sets, supporting the linguistic value of orthographic and phonetic considerations in readability assessment.", }

Current approaches in automatic readability assessment have found success with the use of large language models and transformer architectures. These techniques lead to accuracy improvement, but they do not offer the interpretability that is uniquely required by the audience most often employing readability assessment tools: teachers and educators. Recent work that employs more traditional machine learning methods has highlighted the linguistic importance of considering semantic and syntactic characteristics of text in readability assessment by utilizing handcrafted feature sets. Research in Education suggests that, in addition to semantics and syntax, phonetic and orthographic instruction are necessary for children to progress through the stages of reading and spelling development; children must first learn to decode the letters and symbols on a page to recognize words and phonemes and their connection to speech sounds. Here, we incorporate this word-level phonemic decoding process into readability assessment by crafting a phonetically-based feature set for grade-level classification for English. Our resulting feature set shows comparable performance to much larger, semantically- and syntactically-based feature sets, supporting the linguistic value of orthographic and phonetic considerations in readability assessment.

Incorporating Word-level Phonemic Decoding into Readability Assessment

lrec-main.788

Poster

https://aclanthology.org/2024.lrec-main.789.bib

https://aclanthology.org/2024.lrec-main.789/

@inproceedings{ghosh-etal-2024-indicfinnlp, title = "{I}ndic{F}in{NLP}: Financial Natural Language Processing for {I}ndian Languages", author = "Ghosh, Sohom and Maji, Arnab and Narayana, Aswartha and Naskar, Sudip Kumar", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.789", pages = "9010--9018", abstract = "Applications of Natural Language Processing (NLP) in the finance domain have been very popular of late. For financial NLP, (FinNLP) while various datasets exist for widely spoken languages like English and Chinese, datasets are scarce for low resource languages,particularly for Indian languages. In this paper, we address this challenges by presenting IndicFinNLP {--} a collection of 9 datasets consisting of three tasks relating to FinNLP for three Indian languages. These tasks are Exaggerated Numeral Detection, Sustainability Classification, and ESG Theme Determination of financial texts in Hindi, Bengali, and Telugu. Moreover, we release the datasets under CC BY-NC-SA 4.0 license for the benefit of the research community.", }

Applications of Natural Language Processing (NLP) in the finance domain have been very popular of late. For financial NLP, (FinNLP) while various datasets exist for widely spoken languages like English and Chinese, datasets are scarce for low resource languages,particularly for Indian languages. In this paper, we address this challenges by presenting IndicFinNLP {--} a collection of 9 datasets consisting of three tasks relating to FinNLP for three Indian languages. These tasks are Exaggerated Numeral Detection, Sustainability Classification, and ESG Theme Determination of financial texts in Hindi, Bengali, and Telugu. Moreover, we release the datasets under CC BY-NC-SA 4.0 license for the benefit of the research community.

IndicFinNLP: Financial Natural Language Processing for Indian Languages

lrec-main.789

Poster

https://aclanthology.org/2024.lrec-main.790.bib

https://aclanthology.org/2024.lrec-main.790/

@inproceedings{sethiya-etal-2024-indic, title = "{I}ndic-{TEDST}: Datasets and Baselines for Low-Resource Speech to Text Translation", author = "Sethiya, Nivedita and Nair, Saanvi and Maurya, Chandresh", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.790", pages = "9019--9024", abstract = "Speech-to-text (ST) task is the translation of speech in a language to text in a different language. It has use cases in subtitling, dubbing, etc. Traditionally, ST task has been solved by cascading automatic speech recognition (ASR) and machine translation (MT) models which leads to error propagation, high latency, and training time. To minimize such issues, end-to-end models have been proposed recently. However, we find that only a few works have reported results of ST models on a limited number of low-resource languages. To take a step further in this direction, we release datasets and baselines for low-resource ST tasks. Concretely, our dataset has 9 language pairs and benchmarking has been done against SOTA ST models. The low performance of SOTA ST models on Indic-TEDST data indicates the necessity of the development of ST models specifically designed for low-resource languages.", }

Speech-to-text (ST) task is the translation of speech in a language to text in a different language. It has use cases in subtitling, dubbing, etc. Traditionally, ST task has been solved by cascading automatic speech recognition (ASR) and machine translation (MT) models which leads to error propagation, high latency, and training time. To minimize such issues, end-to-end models have been proposed recently. However, we find that only a few works have reported results of ST models on a limited number of low-resource languages. To take a step further in this direction, we release datasets and baselines for low-resource ST tasks. Concretely, our dataset has 9 language pairs and benchmarking has been done against SOTA ST models. The low performance of SOTA ST models on Indic-TEDST data indicates the necessity of the development of ST models specifically designed for low-resource languages.

Indic-TEDST: Datasets and Baselines for Low-Resource Speech to Text Translation

lrec-main.790

Poster

https://aclanthology.org/2024.lrec-main.791.bib

https://aclanthology.org/2024.lrec-main.791/

@inproceedings{muller-plank-2024-indirectqa, title = "{I}ndirect{QA}: Understanding Indirect Answers to Implicit Polar Questions in {F}rench and {S}panish", author = {M{\"u}ller, Christin and Plank, Barbara}, editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.791", pages = "9025--9035", abstract = "Polar questions are common in dialogue and expect exactly one of two answers (yes/no). It is however not uncommon for speakers to bypass these expected choices and answer, for example, {``}Islands are generally by the sea{''} to the question: {``}An island? By the sea?{''}. While such answers are natural in spoken dialogues, conversational systems still struggle to interpret them. Seminal work to interpret indirect answers were made in recent years{---}but only for English and with strict question formulations. In this work, we present a new corpus for French and Spanish{---}IndirectQA {---}where we mine subtitle data for indirect answers to study the labeling task with six different labels, while broadening polar questions to include also implicit polar questions (statements that trigger a yes/no-answer which are not necessarily formulated as a question). We opted for subtitles since they are a readily available source of conversation in various languages, but also come with peculiarities and challenges which we will discuss. Overall, we provide the first results on French and Spanish. They show that the task is challenging: the baseline accuracy scores drop from 61.43 on English to 44.06 for French and Spanish.", }

Polar questions are common in dialogue and expect exactly one of two answers (yes/no). It is however not uncommon for speakers to bypass these expected choices and answer, for example, {``}Islands are generally by the sea{''} to the question: {``}An island? By the sea?{''}. While such answers are natural in spoken dialogues, conversational systems still struggle to interpret them. Seminal work to interpret indirect answers were made in recent years{---}but only for English and with strict question formulations. In this work, we present a new corpus for French and Spanish{---}IndirectQA {---}where we mine subtitle data for indirect answers to study the labeling task with six different labels, while broadening polar questions to include also implicit polar questions (statements that trigger a yes/no-answer which are not necessarily formulated as a question). We opted for subtitles since they are a readily available source of conversation in various languages, but also come with peculiarities and challenges which we will discuss. Overall, we provide the first results on French and Spanish. They show that the task is challenging: the baseline accuracy scores drop from 61.43 on English to 44.06 for French and Spanish.

IndirectQA: Understanding Indirect Answers to Implicit Polar Questions in French and Spanish

lrec-main.791

Poster

https://aclanthology.org/2024.lrec-main.792.bib

https://aclanthology.org/2024.lrec-main.792/

@inproceedings{anil-etal-2024-inductive, title = "Inductive Knowledge Graph Completion with {GNN}s and Rules: An Analysis", author = "Anil, Akash and Gutierrez-Basulto, Victor and Ibanez-Garcia, Yazmin and Schockaert, Steven", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.792", pages = "9036--9049", abstract = "The task of inductive knowledge graph completion requires models to learn inference patterns from a training graph, which can then be used to make predictions on a disjoint test graph. Rule-based methods seem like a natural fit for this task, but in practice they significantly underperform state-of-the-art methods based on Graph Neural Networks (GNNs), such as NBFNet. We hypothesise that the underperformance of rule-based methods is due to two factors: (i) implausible entities are not ranked at all and (ii) only the most informative path is taken into account when determining the confidence in a given link prediction answer. To analyse the impact of these factors, we study a number of variants of a rule-based approach, which are specifically aimed at addressing the aforementioned issues. We find that the resulting models can achieve a performance which is close to that of NBFNet. Crucially, the considered variants only use a small fraction of the evidence that NBFNet relies on, which means that they largely keep the interpretability advantage of rule-based methods. Moreover, we show that a further variant, which does look at the full KG, consistently outperforms NBFNet.", }

The task of inductive knowledge graph completion requires models to learn inference patterns from a training graph, which can then be used to make predictions on a disjoint test graph. Rule-based methods seem like a natural fit for this task, but in practice they significantly underperform state-of-the-art methods based on Graph Neural Networks (GNNs), such as NBFNet. We hypothesise that the underperformance of rule-based methods is due to two factors: (i) implausible entities are not ranked at all and (ii) only the most informative path is taken into account when determining the confidence in a given link prediction answer. To analyse the impact of these factors, we study a number of variants of a rule-based approach, which are specifically aimed at addressing the aforementioned issues. We find that the resulting models can achieve a performance which is close to that of NBFNet. Crucially, the considered variants only use a small fraction of the evidence that NBFNet relies on, which means that they largely keep the interpretability advantage of rule-based methods. Moreover, we show that a further variant, which does look at the full KG, consistently outperforms NBFNet.

Inductive Knowledge Graph Completion with GNNs and Rules: An Analysis

lrec-main.792

Poster

2308.07942

https://aclanthology.org/2024.lrec-main.793.bib

https://aclanthology.org/2024.lrec-main.793/

@inproceedings{bencke-etal-2024-inferbr, title = "{I}nfer{BR}: A Natural Language Inference Dataset in {P}ortuguese", author = "Bencke, Luciana and Pereira, Francielle Vasconcellos and Santos, Moniele Kunrath and Moreira, Viviane", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.793", pages = "9050--9060", abstract = "Natural Language Inference semantic concepts are central to all aspects of natural language meaning. Portuguese has few NLI-annotated datasets created through automatic translation followed by manual checking. The manual creation of NLI datasets is complex and requires many efforts that are sometimes unavailable. Thus, investments to produce good quality synthetic instances that could be used to train machine learning models for NLI are welcome. This work produced InferBR, an NLI dataset for Portuguese. We relied on a semiautomatic process to generate premises and an automatic process to generate hypotheses. The dataset was manually revised, showing that 97.4{\%} of the sentence pairs had good quality, and nearly 100{\%} of the instances had the correct label assigned. The model trained with InferBR is better at recognizing entailment classes in the other Portuguese datasets than the reverse. Because of its diversity and many unique sentences, InferBR can potentially be further augmented. In addition to the dataset, a key contribution is our proposed generation processes for premises and hypotheses that can easily be adapted to other languages and tasks.", }

Natural Language Inference semantic concepts are central to all aspects of natural language meaning. Portuguese has few NLI-annotated datasets created through automatic translation followed by manual checking. The manual creation of NLI datasets is complex and requires many efforts that are sometimes unavailable. Thus, investments to produce good quality synthetic instances that could be used to train machine learning models for NLI are welcome. This work produced InferBR, an NLI dataset for Portuguese. We relied on a semiautomatic process to generate premises and an automatic process to generate hypotheses. The dataset was manually revised, showing that 97.4{\%} of the sentence pairs had good quality, and nearly 100{\%} of the instances had the correct label assigned. The model trained with InferBR is better at recognizing entailment classes in the other Portuguese datasets than the reverse. Because of its diversity and many unique sentences, InferBR can potentially be further augmented. In addition to the dataset, a key contribution is our proposed generation processes for premises and hypotheses that can easily be adapted to other languages and tasks.

InferBR: A Natural Language Inference Dataset in Portuguese

lrec-main.793

Poster

https://aclanthology.org/2024.lrec-main.794.bib

https://aclanthology.org/2024.lrec-main.794/

@inproceedings{banerjee-etal-2024-inffeed, title = "{I}nf{F}eed: Influence Functions as a Feedback to Improve the Performance of Subjective Tasks", author = "Banerjee, Somnath and Sarkar, Maulindu and Saha, Punyajoy and Mathew, Binny and Mukherjee, Animesh", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.794", pages = "9061--9072", abstract = "Recently, influence functions present an apparatus for achieving explainability for deep neural models by quantifying the perturbation of individual train instances that might impact a test prediction. Our objectives in this paper are twofold. First we incorporate influence functions as a feedback into the model to improve its performance. Second, in a dataset extension exercise, using influence functions to automatically identify data points that have been initially {`}silver{'} annotated by some existing method and need to be cross-checked (and corrected) by annotators to improve the model performance. To meet these objectives, in this paper, we introduce InfFeed, which uses influence functions to compute the influential instances for a target instance. Toward the first objective, we adjust the label of the target instance based on its influencer(s) label. In doing this, InfFeed outperforms the state-of-the-art baselines (including LLMs) by a maximum macro F1-score margin of almost 4{\%} for hate speech classification, 3.5{\%} for stance classification, and 3{\%} for irony and 2{\%} for sarcasm detection. Toward the second objective we show that manually re-annotating only those silver annotated data points in the extension set that have a negative influence can immensely improve the model performance bringing it very close to the scenario where all the data points in the extension set have gold labels. This allows for huge reduction of the number of data points that need to be manually annotated since out of the silver annotated extension dataset, the influence function scheme picks up {\textasciitilde}1/1000 points that need manual correction.", }

Recently, influence functions present an apparatus for achieving explainability for deep neural models by quantifying the perturbation of individual train instances that might impact a test prediction. Our objectives in this paper are twofold. First we incorporate influence functions as a feedback into the model to improve its performance. Second, in a dataset extension exercise, using influence functions to automatically identify data points that have been initially {`}silver{'} annotated by some existing method and need to be cross-checked (and corrected) by annotators to improve the model performance. To meet these objectives, in this paper, we introduce InfFeed, which uses influence functions to compute the influential instances for a target instance. Toward the first objective, we adjust the label of the target instance based on its influencer(s) label. In doing this, InfFeed outperforms the state-of-the-art baselines (including LLMs) by a maximum macro F1-score margin of almost 4{\%} for hate speech classification, 3.5{\%} for stance classification, and 3{\%} for irony and 2{\%} for sarcasm detection. Toward the second objective we show that manually re-annotating only those silver annotated data points in the extension set that have a negative influence can immensely improve the model performance bringing it very close to the scenario where all the data points in the extension set have gold labels. This allows for huge reduction of the number of data points that need to be manually annotated since out of the silver annotated extension dataset, the influence function scheme picks up {\textasciitilde}1/1000 points that need manual correction.

InfFeed: Influence Functions as a Feedback to Improve the Performance of Subjective Tasks

lrec-main.794

Poster

2402.14702

https://aclanthology.org/2024.lrec-main.795.bib

https://aclanthology.org/2024.lrec-main.795/

@inproceedings{xie-etal-2024-infoenh, title = "{I}nfo{E}nh: Towards Multimodal Sentiment Analysis via Information Bottleneck Filter and Optimal Transport Alignment", author = "Xie, Yifeng and Zhu, Zhihong and Lu, Xuan and Huang, Zhiqi and Xiong, Haoran", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.795", pages = "9073--9083", abstract = "In recent years, Multimodal Sentiment Analysis (MSA) leveraging deep learning has demonstrated exceptional performance in a wide range of domains. Its success lies in effectively utilizing information from multiple modalities to analyze sentiments. Despite these advancements, MSA is confronted with two significant challenges. Firstly, each modality often has a surplus of unimportance data, which can overshadow the essential information. Secondly, the crucial cues for sentiment analysis may conflict across different modalities, thereby complicating the analysis process. These issues have a certain impact on the model{'}s effectiveness in MSA tasks. To address these challenges, this paper introduces a novel method tailored for MSA, termed InfoEnh. This approach utilizes a masking technique as the bottleneck for information filtering, simultaneously maximizing mutual information to retain crucial data. Furthermore, the method integrates all modalities into a common feature space via domain adaptation, which is enhanced by the application of optimal transport. Extensive experiments conducted on two benchmark MSA datasets demonstrate the effectiveness of our proposed approach. Further analyzes indicate significant improvements over the baselines.", }

In recent years, Multimodal Sentiment Analysis (MSA) leveraging deep learning has demonstrated exceptional performance in a wide range of domains. Its success lies in effectively utilizing information from multiple modalities to analyze sentiments. Despite these advancements, MSA is confronted with two significant challenges. Firstly, each modality often has a surplus of unimportance data, which can overshadow the essential information. Secondly, the crucial cues for sentiment analysis may conflict across different modalities, thereby complicating the analysis process. These issues have a certain impact on the model{'}s effectiveness in MSA tasks. To address these challenges, this paper introduces a novel method tailored for MSA, termed InfoEnh. This approach utilizes a masking technique as the bottleneck for information filtering, simultaneously maximizing mutual information to retain crucial data. Furthermore, the method integrates all modalities into a common feature space via domain adaptation, which is enhanced by the application of optimal transport. Extensive experiments conducted on two benchmark MSA datasets demonstrate the effectiveness of our proposed approach. Further analyzes indicate significant improvements over the baselines.

InfoEnh: Towards Multimodal Sentiment Analysis via Information Bottleneck Filter and Optimal Transport Alignment

lrec-main.795

Poster

https://aclanthology.org/2024.lrec-main.796.bib

https://aclanthology.org/2024.lrec-main.796/

@inproceedings{el-khbir-etal-2024-information, title = "Information Extraction with Differentiable Beam Search on Graph {RNN}s", author = "El Khbir, Niama and Tomeh, Nadi and Charnois, Thierry", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.796", pages = "9084--9096", abstract = "Information extraction (IE) from text documents is an important NLP task that includes entity, relation, and event extraction. These tasks are often addressed jointly as a graph generation problem, where entities and event triggers represent nodes and where relations and event arguments represent edges. Most existing systems use local classifiers for nodes and edges, trained using cross-entropy loss, and employ inference strategies such as beam search to approximate the optimal graph structure. These approaches typically suffer from exposure bias due to the discrepancy between training and decoding. In this paper, we tackle this problem by casting graph generation as auto-regressive sequence labeling and making its training aware of the decoding procedure by using a differentiable version of beam search. We evaluate the effectiveness of our approach through extensive experiments conducted on the ACE05 and ConLL04 datasets across diverse languages. Our experimental findings affirm that our model outperforms its non-decoding-aware version for all datasets employed. Furthermore, we conduct ablation studies that emphasize the effectiveness of aligning training and inference. Additionally, we introduce a novel quantification of exposure bias within this context, providing valuable insights into the functioning of our model.", }

Information extraction (IE) from text documents is an important NLP task that includes entity, relation, and event extraction. These tasks are often addressed jointly as a graph generation problem, where entities and event triggers represent nodes and where relations and event arguments represent edges. Most existing systems use local classifiers for nodes and edges, trained using cross-entropy loss, and employ inference strategies such as beam search to approximate the optimal graph structure. These approaches typically suffer from exposure bias due to the discrepancy between training and decoding. In this paper, we tackle this problem by casting graph generation as auto-regressive sequence labeling and making its training aware of the decoding procedure by using a differentiable version of beam search. We evaluate the effectiveness of our approach through extensive experiments conducted on the ACE05 and ConLL04 datasets across diverse languages. Our experimental findings affirm that our model outperforms its non-decoding-aware version for all datasets employed. Furthermore, we conduct ablation studies that emphasize the effectiveness of aligning training and inference. Additionally, we introduce a novel quantification of exposure bias within this context, providing valuable insights into the functioning of our model.

Information Extraction with Differentiable Beam Search on Graph RNNs

lrec-main.796

Poster

https://aclanthology.org/2024.lrec-main.797.bib

https://aclanthology.org/2024.lrec-main.797/

@inproceedings{diddee-etal-2024-inmt, title = "{INMT}-Lite: Accelerating Low-Resource Language Data Collection via Offline Interactive Neural Machine Translation", author = "Diddee, Harshita and Shukla, Anurag and Ganu, Tanuja and Seshadri, Vivek and Dandapat, Sandipan and Choudhury, Monojit and Bali, Kalika", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.797", pages = "9097--9109", abstract = "A steady increase in the performance of Massively Multilingual Models (MMLMs) has contributed to their rapidly increasing use in data collection pipelines. Interactive Neural Machine Translation (INMT) systems are one class of tools that can utilize MMLMs to promote such data collection in several under-resourced languages. However, these tools are often not adapted to the deployment constraints that native language speakers operate in, as bloated, online inference-oriented MMLMs trained for data-rich languages, drive them. INMT-Lite addresses these challenges through its support of (1) three different modes of Internet-independent deployment and (2) a suite of four assistive interfaces suitable for (3) data-sparse languages. We perform an extensive user study for INMT-Lite with an under-resourced language community, Gondi, to find that INMT-Lite improves the data generation experience of community members along multiple axes, such as cognitive load, task productivity, and interface interaction time and effort, without compromising on the quality of the generated translations.INMT-Lite{'}s code is open-sourced to further research in this domain.", }

A steady increase in the performance of Massively Multilingual Models (MMLMs) has contributed to their rapidly increasing use in data collection pipelines. Interactive Neural Machine Translation (INMT) systems are one class of tools that can utilize MMLMs to promote such data collection in several under-resourced languages. However, these tools are often not adapted to the deployment constraints that native language speakers operate in, as bloated, online inference-oriented MMLMs trained for data-rich languages, drive them. INMT-Lite addresses these challenges through its support of (1) three different modes of Internet-independent deployment and (2) a suite of four assistive interfaces suitable for (3) data-sparse languages. We perform an extensive user study for INMT-Lite with an under-resourced language community, Gondi, to find that INMT-Lite improves the data generation experience of community members along multiple axes, such as cognitive load, task productivity, and interface interaction time and effort, without compromising on the quality of the generated translations.INMT-Lite{'}s code is open-sourced to further research in this domain.

INMT-Lite: Accelerating Low-Resource Language Data Collection via Offline Interactive Neural Machine Translation

lrec-main.797

Poster

https://aclanthology.org/2024.lrec-main.798.bib

https://aclanthology.org/2024.lrec-main.798/

@inproceedings{tran-miyao-2024-integrating, title = "Integrating Headedness Information into an Auto-generated Multilingual {CCG}bank for Improved Semantic Interpretation", author = "Tran, Tu-Anh and Miyao, Yusuke", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.798", pages = "9110--9119", abstract = "Previously, we introduced a method to generate a multilingual Combinatory Categorial Grammar (CCG) treebank by converting from the Universal Dependencies (UD). However, the method only produces bare CCG derivations without any accompanying semantic representations, which makes it difficult to obtain satisfactory analyses for constructions that involve non-local dependencies, such as control/raising or relative clauses, and limits the general applicability of the treebank. In this work, we present an algorithm that adds semantic representations to existing CCG derivations, in the form of predicate-argument structures. Through hand-crafted rules, we enhance each CCG category with headedness information, with which both local and non-local dependencies can be properly projected. This information is extracted from various sources, including UD, Enhanced UD, and proposition banks. Evaluation of our projected dependencies on the English PropBank and the Universal PropBank 2.0 shows that they can capture most of the semantic dependencies in the target corpora. Further error analysis measures the effectiveness of our algorithm for each language tested, and reveals several issues with the previous method and source data.", }

Previously, we introduced a method to generate a multilingual Combinatory Categorial Grammar (CCG) treebank by converting from the Universal Dependencies (UD). However, the method only produces bare CCG derivations without any accompanying semantic representations, which makes it difficult to obtain satisfactory analyses for constructions that involve non-local dependencies, such as control/raising or relative clauses, and limits the general applicability of the treebank. In this work, we present an algorithm that adds semantic representations to existing CCG derivations, in the form of predicate-argument structures. Through hand-crafted rules, we enhance each CCG category with headedness information, with which both local and non-local dependencies can be properly projected. This information is extracted from various sources, including UD, Enhanced UD, and proposition banks. Evaluation of our projected dependencies on the English PropBank and the Universal PropBank 2.0 shows that they can capture most of the semantic dependencies in the target corpora. Further error analysis measures the effectiveness of our algorithm for each language tested, and reveals several issues with the previous method and source data.

Integrating Headedness Information into an Auto-generated Multilingual CCGbank for Improved Semantic Interpretation

lrec-main.798

Poster

https://aclanthology.org/2024.lrec-main.799.bib

https://aclanthology.org/2024.lrec-main.799/

@inproceedings{du-etal-2024-integrating, title = "Integrating Representation Subspace Mapping with Unimodal Auxiliary Loss for Attention-based Multimodal Emotion Recognition", author = "Du, Xulong and Zhang, Xingnan and Wang, Dandan and Xu, Yingying and Wu, Zhiyuan and Zhang, Shiqing and Zhao, Xiaoming and Yu, Jun and Lou, Liangliang", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.799", pages = "9120--9130", abstract = "Multimodal emotion recognition (MER) aims to identify emotions by utilizing affective information from multiple modalities. Due to the inherent disparities among these heterogeneous modalities, there is a large modality gap in their representations, leading to the challenge of fusing multiple modalities for MER. To address this issue, this work proposes a novel attention-based MER framework by integrating representation subspace mapping with unimodal auxiliary loss for enhancing multimodal fusion capabilities. Initially, a representation subspace mapping module is proposed to map each modality into two distinct subspaces. One is modality-public, enabling the acquisition of common representations and reducing the discrepancies across modalities. The other is modality-unique, retaining the unique characteristics of each modality while eliminating redundant inter-modal attributes. Then, a cross-modality attention is leveraged to bridge the modality gap in unique representations and facilitate modality adaptation. Additionally, our method designs an unimodal auxiliary loss to remove the noise unrelated to emotion classification, resulting in robust and meaningful representations for MER. Comprehensive experiments are conducted on the IEMOCAP and MSP-Improv datasets, and experiment results show that our method achieves superior performance to state-of-the-art MER methods. Keywords: Multimodal emotion recognition, representation subspace mapping, cross-modality attention, unimodal auxiliary loss, fusion", }

Multimodal emotion recognition (MER) aims to identify emotions by utilizing affective information from multiple modalities. Due to the inherent disparities among these heterogeneous modalities, there is a large modality gap in their representations, leading to the challenge of fusing multiple modalities for MER. To address this issue, this work proposes a novel attention-based MER framework by integrating representation subspace mapping with unimodal auxiliary loss for enhancing multimodal fusion capabilities. Initially, a representation subspace mapping module is proposed to map each modality into two distinct subspaces. One is modality-public, enabling the acquisition of common representations and reducing the discrepancies across modalities. The other is modality-unique, retaining the unique characteristics of each modality while eliminating redundant inter-modal attributes. Then, a cross-modality attention is leveraged to bridge the modality gap in unique representations and facilitate modality adaptation. Additionally, our method designs an unimodal auxiliary loss to remove the noise unrelated to emotion classification, resulting in robust and meaningful representations for MER. Comprehensive experiments are conducted on the IEMOCAP and MSP-Improv datasets, and experiment results show that our method achieves superior performance to state-of-the-art MER methods. Keywords: Multimodal emotion recognition, representation subspace mapping, cross-modality attention, unimodal auxiliary loss, fusion

Integrating Representation Subspace Mapping with Unimodal Auxiliary Loss for Attention-based Multimodal Emotion Recognition

lrec-main.799

Poster

https://aclanthology.org/2024.lrec-main.800.bib

https://aclanthology.org/2024.lrec-main.800/

@inproceedings{wang-etal-2024-intent, title = "Intent-Aware and Hate-Mitigating Counterspeech Generation via Dual-Discriminator Guided {LLM}s", author = "Wang, Haiyang and Tian, Zhiliang and Song, Xin and Zhang, Yue and Pan, Yuchen and Tu, Hongkui and Huang, Minlie and Zhou, Bin", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.800", pages = "9131--9142", abstract = "Counterspeech is an effective way to combat online hate speech. Considering the multifaceted nature of online hate speech, counterspeech with varying intents (e.g., denouncing or empathy) has significant potential to mitigate hate speech effectively. Recently, controlled approaches based on large language models (LLMs) have been explored to generate intent-specific counterspeech. Due to the lack of attention to intent-specific information by LLMs during the decoding process, those methods cater more to the semantic information rather than matching with the desired intents. Further, there are still limitations in quantitatively evaluating the effectiveness of counterspeech with different intents in mitigating hate speech. In this paper, to address the above issues, we propose DART, an LLMs-based DuAl-discRiminaTor guided framework for counterspeech generation. We employ an intent-aware discriminator and hate-mitigating discriminator to jointly guide the decoding preferences of LLMs, which facilitates the model towards generating counterspeech catering to specific intent and hate mitigation. We apply a maximum-margin relative objective for training discriminators. This objective leverages the distance between counterspeech aligned with the desired target (such as specific intent or effectiveness in hate mitigation) and undesired as an effective learning signal. Extensive experiments show that DART achieves excellent performances in matching the desired intent and mitigating hate.", }

Counterspeech is an effective way to combat online hate speech. Considering the multifaceted nature of online hate speech, counterspeech with varying intents (e.g., denouncing or empathy) has significant potential to mitigate hate speech effectively. Recently, controlled approaches based on large language models (LLMs) have been explored to generate intent-specific counterspeech. Due to the lack of attention to intent-specific information by LLMs during the decoding process, those methods cater more to the semantic information rather than matching with the desired intents. Further, there are still limitations in quantitatively evaluating the effectiveness of counterspeech with different intents in mitigating hate speech. In this paper, to address the above issues, we propose DART, an LLMs-based DuAl-discRiminaTor guided framework for counterspeech generation. We employ an intent-aware discriminator and hate-mitigating discriminator to jointly guide the decoding preferences of LLMs, which facilitates the model towards generating counterspeech catering to specific intent and hate mitigation. We apply a maximum-margin relative objective for training discriminators. This objective leverages the distance between counterspeech aligned with the desired target (such as specific intent or effectiveness in hate mitigation) and undesired as an effective learning signal. Extensive experiments show that DART achieves excellent performances in matching the desired intent and mitigating hate.

Intent-Aware and Hate-Mitigating Counterspeech Generation via Dual-Discriminator Guided LLMs

lrec-main.800

Poster