Datasets:

EMNLP

/

EMNLP2023-papers

like 0

Follow

Conference on Empirical Methods in Natural Language Processing 1

https://aclanthology.org/2023.findings-emnlp.1020.bib

https://aclanthology.org/2023.findings-emnlp.1020/

@inproceedings{patel-etal-2023-learning, title = "Learning Interpretable Style Embeddings via Prompting {LLM}s", author = "Patel, Ajay and Rao, Delip and Kothary, Ansh and McKeown, Kathleen and Callison-Burch, Chris", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1020", doi = "10.18653/v1/2023.findings-emnlp.1020", pages = "15270--15290", abstract = "Style representation learning builds content-independent representations of author style in text. To date, no large dataset of texts with stylometric annotations on a wide range of style dimensions has been compiled, perhaps because the linguistic expertise to perform such annotation would be prohibitively expensive. Therefore, current style representation approaches make use of unsupervised neural methods to disentangle style from content to create style vectors. These approaches, however, result in uninterpretable representations, complicating their usage in downstream applications like authorship attribution where auditing and explainability is critical. In this work, we use prompting to perform stylometry on a large number of texts to generate a synthetic stylometry dataset. We use this synthetic data to then train human-interpretable style representations we call LISA embeddings. We release our synthetic dataset (StyleGenome) and our interpretable style embedding model (LISA) as resources.", }

Style representation learning builds content-independent representations of author style in text. To date, no large dataset of texts with stylometric annotations on a wide range of style dimensions has been compiled, perhaps because the linguistic expertise to perform such annotation would be prohibitively expensive. Therefore, current style representation approaches make use of unsupervised neural methods to disentangle style from content to create style vectors. These approaches, however, result in uninterpretable representations, complicating their usage in downstream applications like authorship attribution where auditing and explainability is critical. In this work, we use prompting to perform stylometry on a large number of texts to generate a synthetic stylometry dataset. We use this synthetic data to then train human-interpretable style representations we call LISA embeddings. We release our synthetic dataset (StyleGenome) and our interpretable style embedding model (LISA) as resources.

Learning Interpretable Style Embeddings via Prompting LLMs

findings-emnlp.1020

2305.12696

https://huggingface.co/papers/2305.12696

Poster

https://aclanthology.org/2023.findings-emnlp.1021.bib

https://aclanthology.org/2023.findings-emnlp.1021/

@inproceedings{hu-etal-2023-exploring, title = "Exploring Context-Aware Evaluation Metrics for Machine Translation", author = "Hu, Xinyu and Yin, Xunjian and Wan, Xiaojun", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1021", doi = "10.18653/v1/2023.findings-emnlp.1021", pages = "15291--15298", abstract = "Previous studies on machine translation evaluation mostly focused on the quality of individual sentences, while overlooking the important role of contextual information. Although WMT Metrics Shared Tasks have introduced context content into the human annotations of translation evaluation since 2019, the relevant metrics and methods still did not take advantage of the corresponding context. In this paper, we propose a context-aware machine translation evaluation metric called Cont-COMET, built upon the effective COMET framework. Our approach simultaneously considers the preceding and subsequent contexts of the sentence to be evaluated and trains our metric to be aligned with the setting during human annotation. We also introduce a content selection method to extract and utilize the most relevant information. The experiments and evaluation of Cont-COMET on the official test framework from WMT show improvements in both system-level and segment-level assessments.", }

Previous studies on machine translation evaluation mostly focused on the quality of individual sentences, while overlooking the important role of contextual information. Although WMT Metrics Shared Tasks have introduced context content into the human annotations of translation evaluation since 2019, the relevant metrics and methods still did not take advantage of the corresponding context. In this paper, we propose a context-aware machine translation evaluation metric called Cont-COMET, built upon the effective COMET framework. Our approach simultaneously considers the preceding and subsequent contexts of the sentence to be evaluated and trains our metric to be aligned with the setting during human annotation. We also introduce a content selection method to extract and utilize the most relevant information. The experiments and evaluation of Cont-COMET on the official test framework from WMT show improvements in both system-level and segment-level assessments.

Exploring Context-Aware Evaluation Metrics for Machine Translation

findings-emnlp.1021

Poster

https://aclanthology.org/2023.findings-emnlp.1022.bib

https://aclanthology.org/2023.findings-emnlp.1022/

@inproceedings{khalifa-etal-2023-grace, title = "{GRACE}: Discriminator-Guided Chain-of-Thought Reasoning", author = "Khalifa, Muhammad and Logeswaran, Lajanugen and Lee, Moontae and Lee, Honglak and Wang, Lu", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1022", doi = "10.18653/v1/2023.findings-emnlp.1022", pages = "15299--15328", abstract = "In the context of multi-step reasoning, e.g., with chain-of-thought, language models (LMs) can easily assign a high likelihood to incorrect steps. As a result, decoding strategies that optimize for solution likelihood often yield incorrect solutions. To address this issue, we propose Guiding chain-of-thought ReAsoning with a CorrectnEss Discriminator (GRACE), a stepwise decoding approach that steers the decoding process towards producing correct reasoning steps. GRACE employs a discriminator trained with a contrastive loss over correct and incorrect steps, which is used during decoding to score next-step candidates based on their correctness. Importantly, GRACE only requires sampling from the LM, without the need for LM training or fine-tuning. Using models from FLAN-T5 and LLaMA families, we evaluate GRACE over four math and two symbolic reasoning tasks, where it exhibits substantial performance gains compared to greedy decoding, verifiers, and self-consistency in most settings. When further combined with self-consistency, GRACE outperforms all the baselines by sizeable margins. Human and LLM evaluations over GSM8K show that GRACE not only improves the final answer accuracy but also the correctness of the intermediate reasoning.", }

In the context of multi-step reasoning, e.g., with chain-of-thought, language models (LMs) can easily assign a high likelihood to incorrect steps. As a result, decoding strategies that optimize for solution likelihood often yield incorrect solutions. To address this issue, we propose Guiding chain-of-thought ReAsoning with a CorrectnEss Discriminator (GRACE), a stepwise decoding approach that steers the decoding process towards producing correct reasoning steps. GRACE employs a discriminator trained with a contrastive loss over correct and incorrect steps, which is used during decoding to score next-step candidates based on their correctness. Importantly, GRACE only requires sampling from the LM, without the need for LM training or fine-tuning. Using models from FLAN-T5 and LLaMA families, we evaluate GRACE over four math and two symbolic reasoning tasks, where it exhibits substantial performance gains compared to greedy decoding, verifiers, and self-consistency in most settings. When further combined with self-consistency, GRACE outperforms all the baselines by sizeable margins. Human and LLM evaluations over GSM8K show that GRACE not only improves the final answer accuracy but also the correctness of the intermediate reasoning.

GRACE: Discriminator-Guided Chain-of-Thought Reasoning

findings-emnlp.1022

2305.14934

https://huggingface.co/papers/2305.14934

Poster

https://aclanthology.org/2023.findings-emnlp.1023.bib

https://aclanthology.org/2023.findings-emnlp.1023/

@inproceedings{shi-etal-2023-qadynamics, title = "{QADYNAMICS}: Training Dynamics-Driven Synthetic {QA} Diagnostic for Zero-Shot Commonsense Question Answering", author = "Shi, Haochen and Wang, Weiqi and Fang, Tianqing and Xu, Baixuan and Ding, Wenxuan and Liu, Xin and Song, Yangqiu", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1023", doi = "10.18653/v1/2023.findings-emnlp.1023", pages = "15329--15341", abstract = "Zero-shot commonsense Question-Answering (QA) requires models to reason about general situations beyond specific benchmarks. State-of-the-art approaches fine-tune language models on QA pairs constructed from CommonSense Knowledge Bases (CSKBs) to equip the models with more commonsense knowledge in a QA context. However, current QA synthesis protocols may introduce noise from the CSKBs and generate ungrammatical questions and false negative options, which impede the model{'}s ability to generalize. To address these issues, we propose QADYNAMICS, a training dynamics-driven framework for QA diagnostics and refinement. Our approach analyzes the training dynamics of each QA pair at both the question level and option level, discarding machine-detectable artifacts by removing uninformative QA pairs and mislabeled or false-negative options. Extensive experiments demonstrate the effectiveness of our approach, which outperforms all baselines while using only 33{\%} of the synthetic data, even including LLMs such as ChatGPT. Moreover, expert evaluations confirm that our framework significantly improves the quality of QA synthesis. Our code and model checkpoints are available at https://github.com/HKUST-KnowComp/QaDynamics.", }

Zero-shot commonsense Question-Answering (QA) requires models to reason about general situations beyond specific benchmarks. State-of-the-art approaches fine-tune language models on QA pairs constructed from CommonSense Knowledge Bases (CSKBs) to equip the models with more commonsense knowledge in a QA context. However, current QA synthesis protocols may introduce noise from the CSKBs and generate ungrammatical questions and false negative options, which impede the model{'}s ability to generalize. To address these issues, we propose QADYNAMICS, a training dynamics-driven framework for QA diagnostics and refinement. Our approach analyzes the training dynamics of each QA pair at both the question level and option level, discarding machine-detectable artifacts by removing uninformative QA pairs and mislabeled or false-negative options. Extensive experiments demonstrate the effectiveness of our approach, which outperforms all baselines while using only 33{\%} of the synthetic data, even including LLMs such as ChatGPT. Moreover, expert evaluations confirm that our framework significantly improves the quality of QA synthesis. Our code and model checkpoints are available at https://github.com/HKUST-KnowComp/QaDynamics.

QADYNAMICS: Training Dynamics-Driven Synthetic QA Diagnostic for Zero-Shot Commonsense Question Answering

findings-emnlp.1023

2310.11303

Poster

https://aclanthology.org/2023.findings-emnlp.1024.bib

https://aclanthology.org/2023.findings-emnlp.1024/

@inproceedings{liu-etal-2023-rexuie, title = "{R}ex{UIE}: A Recursive Method with Explicit Schema Instructor for Universal Information Extraction", author = "Liu, Chengyuan and Zhao, Fubang and Kang, Yangyang and Zhang, Jingyuan and Zhou, Xiang and Sun, Changlong and Kuang, Kun and Wu, Fei", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1024", doi = "10.18653/v1/2023.findings-emnlp.1024", pages = "15342--15359", abstract = "Universal Information Extraction (UIE) is an area of interest due to the challenges posed by varying targets, heterogeneous structures, and demand-specific schemas. Previous works have achieved success by unifying a few tasks, such as Named Entity Recognition (NER) and Relation Extraction (RE), while they fall short of being true UIE models particularly when extracting other general schemas such as quadruples and quintuples. Additionally, these models used an implicit structural schema instructor, which could lead to incorrect links between types, hindering the model{'}s generalization and performance in low-resource scenarios. In this paper, we redefine the true UIE with a formal formulation that covers almost all extraction schemas. To the best of our knowledge, we are the first to introduce UIE for any kind of schemas. In addition, we propose RexUIE, which is a Recursive Method with Explicit Schema Instructor for UIE. To avoid interference between different types, we reset the position ids and attention mask matrices. RexUIE shows strong performance under both full-shot and few-shot settings and achieves state-of-the-art results on the tasks of extracting complex schemas.", }

Universal Information Extraction (UIE) is an area of interest due to the challenges posed by varying targets, heterogeneous structures, and demand-specific schemas. Previous works have achieved success by unifying a few tasks, such as Named Entity Recognition (NER) and Relation Extraction (RE), while they fall short of being true UIE models particularly when extracting other general schemas such as quadruples and quintuples. Additionally, these models used an implicit structural schema instructor, which could lead to incorrect links between types, hindering the model{'}s generalization and performance in low-resource scenarios. In this paper, we redefine the true UIE with a formal formulation that covers almost all extraction schemas. To the best of our knowledge, we are the first to introduce UIE for any kind of schemas. In addition, we propose RexUIE, which is a Recursive Method with Explicit Schema Instructor for UIE. To avoid interference between different types, we reset the position ids and attention mask matrices. RexUIE shows strong performance under both full-shot and few-shot settings and achieves state-of-the-art results on the tasks of extracting complex schemas.

RexUIE: A Recursive Method with Explicit Schema Instructor for Universal Information Extraction

findings-emnlp.1024

2304.14770

Poster

https://aclanthology.org/2023.findings-emnlp.1025.bib

https://aclanthology.org/2023.findings-emnlp.1025/

@inproceedings{zeng-etal-2023-promptara, title = "{P}rompt{ARA}: Improving Deep Representation in Hybrid Automatic Readability Assessment with Prompt and Orthogonal Projection", author = "Zeng, Jinshan and Yu, Xianglong and Tong, Xianchao and Xiao, Wenyan", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1025", doi = "10.18653/v1/2023.findings-emnlp.1025", pages = "15360--15371", abstract = "Readability assessment aims to automatically classify texts based on readers{'} reading levels. The hybrid automatic readability assessment (ARA) models using both deep and linguistic features have attracted rising attention in recent years due to their impressive performance. However, deep features are not fully explored due to the scarcity of training data, and the fusion of deep and linguistic features is not very effective in existing hybrid ARA models. In this paper, we propose a novel hybrid ARA model called PromptARA through employing prompts to improve deep feature representations and an orthogonal projection layer to fuse both deep and linguistic features. A series of experiments are conducted over four English and two Chinese corpora to show the effectiveness of the proposed model. Experimental results demonstrate that the proposed model is superior to state-of-the-art models.", }

Readability assessment aims to automatically classify texts based on readers{'} reading levels. The hybrid automatic readability assessment (ARA) models using both deep and linguistic features have attracted rising attention in recent years due to their impressive performance. However, deep features are not fully explored due to the scarcity of training data, and the fusion of deep and linguistic features is not very effective in existing hybrid ARA models. In this paper, we propose a novel hybrid ARA model called PromptARA through employing prompts to improve deep feature representations and an orthogonal projection layer to fuse both deep and linguistic features. A series of experiments are conducted over four English and two Chinese corpora to show the effectiveness of the proposed model. Experimental results demonstrate that the proposed model is superior to state-of-the-art models.

PromptARA: Improving Deep Representation in Hybrid Automatic Readability Assessment with Prompt and Orthogonal Projection

findings-emnlp.1025

Poster

https://aclanthology.org/2023.findings-emnlp.1026.bib

https://aclanthology.org/2023.findings-emnlp.1026/

@inproceedings{wang-buschmeier-2023-listener, title = "Does Listener Gaze in Face-to-Face Interaction Follow the Entropy Rate Constancy Principle: An Empirical Study", author = "Wang, Yu and Buschmeier, Hendrik", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1026", doi = "10.18653/v1/2023.findings-emnlp.1026", pages = "15372--15379", abstract = "It is generally assumed that language (written and spoken) follows the entropy rate constancy (ERC) principle, which states that the information density of a text is constant over time. Recently, this has also been found for nonverbal gestures used in monologue, but it is still unclear whether the ERC principle also applies to listeners{'} nonverbal signals. We focus on listeners{'} gaze behaviour extracted from video-recorded conversations and trained a transformer-based neural sequence model to process the gaze data of the dialogues and compute its information density. We also compute the information density of the corresponding speech using a pre-trained language model. Our results show (1) that listeners{'} gaze behaviour in dialogues roughly follows the ERC principle, as well as (2) a congruence between information density of speech and listeners{'} gaze behaviour.", }

It is generally assumed that language (written and spoken) follows the entropy rate constancy (ERC) principle, which states that the information density of a text is constant over time. Recently, this has also been found for nonverbal gestures used in monologue, but it is still unclear whether the ERC principle also applies to listeners{'} nonverbal signals. We focus on listeners{'} gaze behaviour extracted from video-recorded conversations and trained a transformer-based neural sequence model to process the gaze data of the dialogues and compute its information density. We also compute the information density of the corresponding speech using a pre-trained language model. Our results show (1) that listeners{'} gaze behaviour in dialogues roughly follows the ERC principle, as well as (2) a congruence between information density of speech and listeners{'} gaze behaviour.

Does Listener Gaze in Face-to-Face Interaction Follow the Entropy Rate Constancy Principle: An Empirical Study

findings-emnlp.1026

Poster

https://aclanthology.org/2023.findings-emnlp.1027.bib

https://aclanthology.org/2023.findings-emnlp.1027/

@inproceedings{zhang-etal-2023-incorporating-object, title = "Incorporating Object-Level Visual Context for Multimodal Fine-Grained Entity Typing", author = "Zhang, Ying and Fan, Wenbo and Song, Kehui and Zhao, Yu and Sui, Xuhui and Yuan, Xiaojie", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1027", doi = "10.18653/v1/2023.findings-emnlp.1027", pages = "15380--15390", abstract = "Fine-grained entity typing (FGET) aims to assign appropriate fine-grained types to entity mentions within their context, which is an important foundational task in natural language processing. Previous approaches for FGET only utilized textual context information. However, in the form of short text, the contextual semantic information is often insufficient for FGET. In many real-world scenarios, text is often accompanied by images, and the visual context is valuable for FGET. To this end, we firstly propose a new task called multimodal fine-grained entity typing (MFGET). Then we construct a large-scale dataset for multimodal fine-grained entity typing called MFIGER based on FIGER. To fully leverage both textual and visual information, we propose a novel Multimodal Object-Level Visual Context Network (MOVCNet). MOVCNet can capture fine-grained semantic information by detecting objects in images, and effectively merge both textual and visual context. Experimental results demonstrate that our approach achieves superior classification performance compared to previous text-based approaches.", }

Fine-grained entity typing (FGET) aims to assign appropriate fine-grained types to entity mentions within their context, which is an important foundational task in natural language processing. Previous approaches for FGET only utilized textual context information. However, in the form of short text, the contextual semantic information is often insufficient for FGET. In many real-world scenarios, text is often accompanied by images, and the visual context is valuable for FGET. To this end, we firstly propose a new task called multimodal fine-grained entity typing (MFGET). Then we construct a large-scale dataset for multimodal fine-grained entity typing called MFIGER based on FIGER. To fully leverage both textual and visual information, we propose a novel Multimodal Object-Level Visual Context Network (MOVCNet). MOVCNet can capture fine-grained semantic information by detecting objects in images, and effectively merge both textual and visual context. Experimental results demonstrate that our approach achieves superior classification performance compared to previous text-based approaches.

Incorporating Object-Level Visual Context for Multimodal Fine-Grained Entity Typing

findings-emnlp.1027

Poster

https://aclanthology.org/2023.findings-emnlp.1028.bib

https://aclanthology.org/2023.findings-emnlp.1028/

@inproceedings{akhtar-etal-2023-exploring, title = "Exploring the Numerical Reasoning Capabilities of Language Models: A Comprehensive Analysis on Tabular Data", author = "Akhtar, Mubashara and Shankarampeta, Abhilash and Gupta, Vivek and Patil, Arpit and Cocarascu, Oana and Simperl, Elena", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1028", doi = "10.18653/v1/2023.findings-emnlp.1028", pages = "15391--15405", abstract = "Numerical data plays a crucial role in various real-world domains like finance, economics, and science. Thus, understanding and reasoning with numbers are essential in these fields. Recent benchmarks have assessed the numerical reasoning abilities of language models, revealing their limitations in limited and specific numerical aspects. In this paper, we propose a complete hierarchical taxonomy for numerical reasoning skills, encompassing over ten reasoning types across four levels: representation, number sense, manipulation, and complex reasoning. We conduct a comprehensive evaluation of state-of-the-art models on all reasoning types. To identify challenging reasoning types for different model types, we develop a diverse and extensive set of numerical probes and measure performance shifts. By employing a semi-automated approach, we focus on the tabular Natural Language Inference (TNLI) task as a case study. While no single model excels in all reasoning types, FlanT5 (few-/zero-shot) and GPT3.5 (few-shot) demonstrate strong overall numerical reasoning skills compared to other models in our probes.", }

Numerical data plays a crucial role in various real-world domains like finance, economics, and science. Thus, understanding and reasoning with numbers are essential in these fields. Recent benchmarks have assessed the numerical reasoning abilities of language models, revealing their limitations in limited and specific numerical aspects. In this paper, we propose a complete hierarchical taxonomy for numerical reasoning skills, encompassing over ten reasoning types across four levels: representation, number sense, manipulation, and complex reasoning. We conduct a comprehensive evaluation of state-of-the-art models on all reasoning types. To identify challenging reasoning types for different model types, we develop a diverse and extensive set of numerical probes and measure performance shifts. By employing a semi-automated approach, we focus on the tabular Natural Language Inference (TNLI) task as a case study. While no single model excels in all reasoning types, FlanT5 (few-/zero-shot) and GPT3.5 (few-shot) demonstrate strong overall numerical reasoning skills compared to other models in our probes.

Exploring the Numerical Reasoning Capabilities of Language Models: A Comprehensive Analysis on Tabular Data

findings-emnlp.1028

2311.02216

Poster

https://aclanthology.org/2023.findings-emnlp.1029.bib

https://aclanthology.org/2023.findings-emnlp.1029/

@inproceedings{tang-etal-2023-assessing, title = "Assessing Privacy Risks in Language Models: A Case Study on Summarization Tasks", author = "Tang, Ruixiang and Lueck, Gord and Quispe, Rodolfo and Inan, Huseyin and Kulkarni, Janardhan and Hu, Xia", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1029", doi = "10.18653/v1/2023.findings-emnlp.1029", pages = "15406--15418", abstract = "Large language models have revolutionized the field of NLP by achieving state-of-the-art performance on various tasks. However, there is a concern that these models may disclose information in the training data. In this study, we focus on the summarization task and investigate the membership inference (MI) attack: given a sample and black-box access to a model{'}s API, it is possible to determine if the sample was part of the training data. We exploit text similarity and the model{'}s resistance to document modifications as potential MI signals and evaluate their effectiveness on widely used datasets. Our results demonstrate that summarization models are at risk of exposing data membership, even in cases where the reference summary is not available. Furthermore, we discuss several safeguards for training summarization models to protect against MI attacks and discuss the inherent trade-off between privacy and utility.", }

Large language models have revolutionized the field of NLP by achieving state-of-the-art performance on various tasks. However, there is a concern that these models may disclose information in the training data. In this study, we focus on the summarization task and investigate the membership inference (MI) attack: given a sample and black-box access to a model{'}s API, it is possible to determine if the sample was part of the training data. We exploit text similarity and the model{'}s resistance to document modifications as potential MI signals and evaluate their effectiveness on widely used datasets. Our results demonstrate that summarization models are at risk of exposing data membership, even in cases where the reference summary is not available. Furthermore, we discuss several safeguards for training summarization models to protect against MI attacks and discuss the inherent trade-off between privacy and utility.

Assessing Privacy Risks in Language Models: A Case Study on Summarization Tasks

findings-emnlp.1029

2310.13291

Poster

https://aclanthology.org/2023.findings-emnlp.1030.bib

https://aclanthology.org/2023.findings-emnlp.1030/

@inproceedings{hosseini-etal-2023-bert, title = "{BERT} Has More to Offer: {BERT} Layers Combination Yields Better Sentence Embeddings", author = "Hosseini, MohammadSaleh and Munia, Munawara and Khan, Latifur", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1030", doi = "10.18653/v1/2023.findings-emnlp.1030", pages = "15419--15431", abstract = "Obtaining sentence representations from BERT-based models as feature extractors is invaluable as it takes much less time to pre-compute a one-time representation of the data and then use it for the downstream tasks, rather than fine-tune the whole BERT. Most previous works acquire a sentence{'}s representation by passing it to BERT and averaging its last layer. In this paper, we propose that the combination of certain layers of a BERT-based model rested on the data set and model can achieve substantially better results. We empirically show the effectiveness of our method for different BERT-based models on different tasks and data sets. Specifically, on seven standard semantic textual similarity data sets, we outperform the baseline BERT by improving the Spearman{'}s correlation by up to 25.75{\%} and on average 16.32{\%} without any further training. We also achieved state-of-the-art results on eight transfer data sets by reducing the relative error by up to 37.41{\%} and on average 17.92{\%}.", }

Obtaining sentence representations from BERT-based models as feature extractors is invaluable as it takes much less time to pre-compute a one-time representation of the data and then use it for the downstream tasks, rather than fine-tune the whole BERT. Most previous works acquire a sentence{'}s representation by passing it to BERT and averaging its last layer. In this paper, we propose that the combination of certain layers of a BERT-based model rested on the data set and model can achieve substantially better results. We empirically show the effectiveness of our method for different BERT-based models on different tasks and data sets. Specifically, on seven standard semantic textual similarity data sets, we outperform the baseline BERT by improving the Spearman{'}s correlation by up to 25.75{\%} and on average 16.32{\%} without any further training. We also achieved state-of-the-art results on eight transfer data sets by reducing the relative error by up to 37.41{\%} and on average 17.92{\%}.

BERT Has More to Offer: BERT Layers Combination Yields Better Sentence Embeddings

findings-emnlp.1030

Poster

https://aclanthology.org/2023.findings-emnlp.1031.bib

https://aclanthology.org/2023.findings-emnlp.1031/

@inproceedings{wu-etal-2023-extrapolating, title = "Extrapolating Multilingual Understanding Models as Multilingual Generators", author = "Wu, Bohong and Yuan, Fei and Zhao, Hai and Li, Lei and Xu, Jingjing", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1031", doi = "10.18653/v1/2023.findings-emnlp.1031", pages = "15432--15444", abstract = "Multilingual understanding models (or encoder-based), pre-trained via masked language modeling, have achieved promising results on many language understanding tasks (e.g., mBERT). However, these models are not capable of generating high-quality text compared with decoder-based causal language models. Can we transform a pre-trained language understanding model into an effective language generation model? We propose a Semantic-Guided Alignment-then-Denoising (SGA) approach to adapt a multilingual encoder to a multilingual generator with a small number of additional parameters. Experiments show that the proposed approach is an effective adaption method, outperforming widely-used initialization-based methods with gains of 9.4 BLEU on machine translation, 8.1 Rouge-L on question generation, and 5.5 METEOR on story generation on XLM-R$_{large}$. On the other hand, we observe that XLM-R is still inferior to mBART in supervised settings despite better results on zero-shot settings, indicating that more exploration is required to make understanding models strong generators. Our code is available at https://github.com/chengzhipanpan/XLMR4MT.", }

Multilingual understanding models (or encoder-based), pre-trained via masked language modeling, have achieved promising results on many language understanding tasks (e.g., mBERT). However, these models are not capable of generating high-quality text compared with decoder-based causal language models. Can we transform a pre-trained language understanding model into an effective language generation model? We propose a Semantic-Guided Alignment-then-Denoising (SGA) approach to adapt a multilingual encoder to a multilingual generator with a small number of additional parameters. Experiments show that the proposed approach is an effective adaption method, outperforming widely-used initialization-based methods with gains of 9.4 BLEU on machine translation, 8.1 Rouge-L on question generation, and 5.5 METEOR on story generation on XLM-R$_{large}$. On the other hand, we observe that XLM-R is still inferior to mBART in supervised settings despite better results on zero-shot settings, indicating that more exploration is required to make understanding models strong generators. Our code is available at https://github.com/chengzhipanpan/XLMR4MT.

Extrapolating Multilingual Understanding Models as Multilingual Generators

findings-emnlp.1031

2305.13140

https://huggingface.co/papers/2305.13140

Poster

https://aclanthology.org/2023.findings-emnlp.1032.bib

https://aclanthology.org/2023.findings-emnlp.1032/

@inproceedings{zhang-etal-2023-sac3, title = "{SAC}$^3$: Reliable Hallucination Detection in Black-Box Language Models via Semantic-aware Cross-check Consistency", author = "Zhang, Jiaxin and Li, Zhuohang and Das, Kamalika and Malin, Bradley and Kumar, Sricharan", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1032", doi = "10.18653/v1/2023.findings-emnlp.1032", pages = "15445--15458", abstract = "Hallucination detection is a critical step toward understanding the trustworthiness of modern language models (LMs). To achieve this goal, we re-examine existing detection approaches based on the self-consistency of LMs and uncover two types of hallucinations resulting from 1) question-level and 2) model-level, which cannot be effectively identified through self-consistency check alone. Building upon this discovery, we propose a novel sampling-based method, i.e., semantic-aware cross-check consistency (SAC$^3$) that expands on the principle of self-consistency checking. Our SAC$^3$ approach incorporates additional mechanisms to detect both question-level and model-level hallucinations by leveraging advances including semantically equivalent question perturbation and cross-model response consistency checking. Through extensive and systematic empirical analysis, we demonstrate that SAC$^3$ outperforms the state of the art in detecting both non-factual and factual statements across multiple question-answering and open-domain generation benchmarks.", }

Hallucination detection is a critical step toward understanding the trustworthiness of modern language models (LMs). To achieve this goal, we re-examine existing detection approaches based on the self-consistency of LMs and uncover two types of hallucinations resulting from 1) question-level and 2) model-level, which cannot be effectively identified through self-consistency check alone. Building upon this discovery, we propose a novel sampling-based method, i.e., semantic-aware cross-check consistency (SAC$^3$) that expands on the principle of self-consistency checking. Our SAC$^3$ approach incorporates additional mechanisms to detect both question-level and model-level hallucinations by leveraging advances including semantically equivalent question perturbation and cross-model response consistency checking. Through extensive and systematic empirical analysis, we demonstrate that SAC$^3$ outperforms the state of the art in detecting both non-factual and factual statements across multiple question-answering and open-domain generation benchmarks.

SAC^3: Reliable Hallucination Detection in Black-Box Language Models via Semantic-aware Cross-check Consistency

findings-emnlp.1032

Poster

https://aclanthology.org/2023.findings-emnlp.1033.bib

https://aclanthology.org/2023.findings-emnlp.1033/

@inproceedings{jeong-etal-2023-test, title = "Test-Time Self-Adaptive Small Language Models for Question Answering", author = "Jeong, Soyeong and Baek, Jinheon and Cho, Sukmin and Hwang, Sung and Park, Jong", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1033", doi = "10.18653/v1/2023.findings-emnlp.1033", pages = "15459--15469", abstract = "Recent instruction-finetuned large language models (LMs) have achieved notable performances in various tasks, such as question-answering (QA). However, despite their ability to memorize a vast amount of general knowledge across diverse tasks, they might be suboptimal on specific tasks due to their limited capacity to transfer and adapt knowledge to target tasks. Moreover, further finetuning LMs with labeled datasets is often infeasible due to their absence, but it is also questionable if we can transfer smaller LMs having limited knowledge only with unlabeled test data. In this work, we show and investigate the capabilities of smaller self-adaptive LMs, only with unlabeled test data. In particular, we first stochastically generate multiple answers, and then ensemble them while filtering out low-quality samples to mitigate noise from inaccurate labels. Our proposed self-adaption strategy demonstrates significant performance improvements on benchmark QA datasets with higher robustness across diverse prompts, enabling LMs to stay stable. Code is available at: https://github.com/starsuzi/T-SAS.", }

Recent instruction-finetuned large language models (LMs) have achieved notable performances in various tasks, such as question-answering (QA). However, despite their ability to memorize a vast amount of general knowledge across diverse tasks, they might be suboptimal on specific tasks due to their limited capacity to transfer and adapt knowledge to target tasks. Moreover, further finetuning LMs with labeled datasets is often infeasible due to their absence, but it is also questionable if we can transfer smaller LMs having limited knowledge only with unlabeled test data. In this work, we show and investigate the capabilities of smaller self-adaptive LMs, only with unlabeled test data. In particular, we first stochastically generate multiple answers, and then ensemble them while filtering out low-quality samples to mitigate noise from inaccurate labels. Our proposed self-adaption strategy demonstrates significant performance improvements on benchmark QA datasets with higher robustness across diverse prompts, enabling LMs to stay stable. Code is available at: https://github.com/starsuzi/T-SAS.

Test-Time Self-Adaptive Small Language Models for Question Answering

findings-emnlp.1033

2310.13307

https://huggingface.co/papers/2310.13307

Poster

https://aclanthology.org/2023.findings-emnlp.1034.bib

https://aclanthology.org/2023.findings-emnlp.1034/

@inproceedings{sun-etal-2023-expnote, title = "{E}xp{N}ote: Black-box Large Language Models are better Task Solvers with Experience Notebook", author = "Sun, Wangtao and Yu, Xuanqing and He, Shizhu and Zhao, Jun and Liu, Kang", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1034", doi = "10.18653/v1/2023.findings-emnlp.1034", pages = "15470--15481", abstract = "Black-box Large Language Models (LLMs) have shown great power in solving various tasks and are considered general problem solvers. However, LLMs still fail in many specific tasks although understand the task instruction. In this paper, we focus on the problem of boosting the ability of black-box LLMs to solve downstream tasks. We propose ExpNote, an automated framework to help LLMs better adapt to unfamiliar tasks through reflecting and noting experiences from training data and retrieving them from external memory during testing. We evaluate ExpNote on multiple tasks and the experimental results demonstrate that the proposed method significantly improves the performance of black-box LLMs. The data and code are available at https://github.com/forangel2014/ExpNote.", }

Black-box Large Language Models (LLMs) have shown great power in solving various tasks and are considered general problem solvers. However, LLMs still fail in many specific tasks although understand the task instruction. In this paper, we focus on the problem of boosting the ability of black-box LLMs to solve downstream tasks. We propose ExpNote, an automated framework to help LLMs better adapt to unfamiliar tasks through reflecting and noting experiences from training data and retrieving them from external memory during testing. We evaluate ExpNote on multiple tasks and the experimental results demonstrate that the proposed method significantly improves the performance of black-box LLMs. The data and code are available at https://github.com/forangel2014/ExpNote.

ExpNote: Black-box Large Language Models are better Task Solvers with Experience Notebook

findings-emnlp.1034

2311.07032

Poster

https://aclanthology.org/2023.findings-emnlp.1035.bib

https://aclanthology.org/2023.findings-emnlp.1035/

@inproceedings{olariu-etal-2023-evaluating, title = "Evaluating Parameter-Efficient Finetuning Approaches for Pre-trained Models on the Financial Domain", author = "Olariu, Isabella and Lothritz, Cedric and Klein, Jacques and Bissyand{\'e}, Tegawend{\'e} and Guo, Siwen and Haddadan, Shohreh", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1035", doi = "10.18653/v1/2023.findings-emnlp.1035", pages = "15482--15491", abstract = "Large-scale language models with millions, billions, or trillions of trainable parameters are becoming increasingly popular. However, they risk becoming rapidly over-parameterized and the adaptation cost of fully fine-tuning them increases significantly. Storing them becomes progressively impractical as it requires keeping a separate copy of all the fine-tuned weights for each task. By freezing all pre-trained weights during fine-tuning, parameter-efficient tuning approaches have become an appealing alternative to traditional fine-tuning. The performance of these approaches has been evaluated on common NLP tasks of the GLUE benchmark and shown to match full fine-tuning performance, however, their impact is less researched in domain-specific fields such as finance. This work compares the performance of a set of financial BERT-like models to their fully fine-tuned counterparts by leveraging different parameter-efficient tuning methods. We see that results are comparable to traditional fine-tuning while gaining in time and resource efficiency.", }

Large-scale language models with millions, billions, or trillions of trainable parameters are becoming increasingly popular. However, they risk becoming rapidly over-parameterized and the adaptation cost of fully fine-tuning them increases significantly. Storing them becomes progressively impractical as it requires keeping a separate copy of all the fine-tuned weights for each task. By freezing all pre-trained weights during fine-tuning, parameter-efficient tuning approaches have become an appealing alternative to traditional fine-tuning. The performance of these approaches has been evaluated on common NLP tasks of the GLUE benchmark and shown to match full fine-tuning performance, however, their impact is less researched in domain-specific fields such as finance. This work compares the performance of a set of financial BERT-like models to their fully fine-tuned counterparts by leveraging different parameter-efficient tuning methods. We see that results are comparable to traditional fine-tuning while gaining in time and resource efficiency.

Evaluating Parameter-Efficient Finetuning Approaches for Pre-trained Models on the Financial Domain

findings-emnlp.1035

Poster

https://aclanthology.org/2023.findings-emnlp.1036.bib

https://aclanthology.org/2023.findings-emnlp.1036/

@inproceedings{behnamghader-etal-2023-retriever, title = "Can Retriever-Augmented Language Models Reason? The Blame Game Between the Retriever and the Language Model", author = "BehnamGhader, Parishad and Miret, Santiago and Reddy, Siva", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1036", doi = "10.18653/v1/2023.findings-emnlp.1036", pages = "15492--15509", abstract = "Augmenting pretrained language models with retrievers has shown promise in effectively solving common NLP problems, such as language modeling and question answering. In this paper, we evaluate the strengths and weaknesses of popular retriever-augmented language models, namely kNN-LM, REALM, DPR + FiD, Contriever + ATLAS, and Contriever + Flan-T5, in reasoning over retrieved statements across different tasks. Our findings indicate that the simple similarity metric employed by retrievers is insufficient for retrieving all the necessary statements for reasoning. Additionally, the language models do not exhibit strong reasoning even when provided with only the required statements. Furthermore, when combined with imperfect retrievers, the performance of the language models becomes even worse, e.g., Flan-T5{'}s performance drops by 28.6{\%} when retrieving 5 statements using Contriever. While larger language models improve performance, there is still a substantial room for enhancement. Our further analysis indicates that multihop retrieve-and-read is promising for large language models like GPT-3.5, but does not generalize to other language models like Flan-T5-xxl. The code is available at https://github.com/McGill-NLP/retriever-lm-reasoning.", }

Augmenting pretrained language models with retrievers has shown promise in effectively solving common NLP problems, such as language modeling and question answering. In this paper, we evaluate the strengths and weaknesses of popular retriever-augmented language models, namely kNN-LM, REALM, DPR + FiD, Contriever + ATLAS, and Contriever + Flan-T5, in reasoning over retrieved statements across different tasks. Our findings indicate that the simple similarity metric employed by retrievers is insufficient for retrieving all the necessary statements for reasoning. Additionally, the language models do not exhibit strong reasoning even when provided with only the required statements. Furthermore, when combined with imperfect retrievers, the performance of the language models becomes even worse, e.g., Flan-T5{'}s performance drops by 28.6{\%} when retrieving 5 statements using Contriever. While larger language models improve performance, there is still a substantial room for enhancement. Our further analysis indicates that multihop retrieve-and-read is promising for large language models like GPT-3.5, but does not generalize to other language models like Flan-T5-xxl. The code is available at https://github.com/McGill-NLP/retriever-lm-reasoning.

Can Retriever-Augmented Language Models Reason? The Blame Game Between the Retriever and the Language Model

findings-emnlp.1036

2212.09146

https://huggingface.co/papers/2212.09146

Poster

https://aclanthology.org/2023.findings-emnlp.1037.bib

https://aclanthology.org/2023.findings-emnlp.1037/

@inproceedings{srivastava-chiang-2023-bertwich, title = "{BERT}wich: Extending {BERT}{'}s Capabilities to Model Dialectal and Noisy Text", author = "Srivastava, Aarohi and Chiang, David", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1037", doi = "10.18653/v1/2023.findings-emnlp.1037", pages = "15510--15521", abstract = "Real-world NLP applications often deal with nonstandard text (e.g., dialectal, informal, or misspelled text). However, language models like BERT deteriorate in the face of dialect variation or noise. How do we push BERT{'}s modeling capabilities to encompass nonstandard text? Fine-tuning helps, but it is designed for specializing a model to a task and does not seem to bring about the deeper, more pervasive changes needed to adapt a model to nonstandard language. In this paper, we introduce the novel idea of sandwiching BERT{'}s encoder stack between additional encoder layers trained to perform masked language modeling on noisy text. We find that our approach, paired with recent work on including character-level noise in fine-tuning data, can promote zero-shot transfer to dialectal text, as well as reduce the distance in the embedding space between words and their noisy counterparts.", }

Real-world NLP applications often deal with nonstandard text (e.g., dialectal, informal, or misspelled text). However, language models like BERT deteriorate in the face of dialect variation or noise. How do we push BERT{'}s modeling capabilities to encompass nonstandard text? Fine-tuning helps, but it is designed for specializing a model to a task and does not seem to bring about the deeper, more pervasive changes needed to adapt a model to nonstandard language. In this paper, we introduce the novel idea of sandwiching BERT{'}s encoder stack between additional encoder layers trained to perform masked language modeling on noisy text. We find that our approach, paired with recent work on including character-level noise in fine-tuning data, can promote zero-shot transfer to dialectal text, as well as reduce the distance in the embedding space between words and their noisy counterparts.

BERTwich: Extending BERT's Capabilities to Model Dialectal and Noisy Text

findings-emnlp.1037

2311.00116

Poster

https://aclanthology.org/2023.findings-emnlp.1038.bib

https://aclanthology.org/2023.findings-emnlp.1038/

@inproceedings{zhou-etal-2023-closed, title = "Closed Boundary Learning for Classification Tasks with the Universum Class", author = "Zhou, Hanzhang and Feng, Zijian and Mao, Kezhi", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1038", doi = "10.18653/v1/2023.findings-emnlp.1038", pages = "15522--15536", abstract = "The Universum class, often known as the *other* class or the*miscellaneous* class, is defined as a collection of samples that do not belong to any class of interest. It is a typical class that exists in many classification-based tasks in NLP, such as relation extraction, named entity recognition, sentiment analysis, etc. The Universum class exhibits very different properties, namely heterogeneity and lack of representativeness in training data; however, existing methods often treat the Universum class equally with the classes of interest, leading to problems such as overfitting, misclassification, and diminished model robustness. In this work, we propose a closed boundary learning method that applies closed decision boundaries to classes of interest and designates the area outside all closed boundaries in the feature space as the space of the Universum class. Specifically, we formulate closed boundaries as arbitrary shapes, propose the inter-class rule-based probability estimation for the Universum class to cater to its unique properties, and propose a boundary learning loss to adjust decision boundaries based on the balance of misclassified samples inside and outside the boundary. In adherence to the natural properties of the Universum class, our method enhances both accuracy and robustness of classification models, demonstrated by improvements on six state-of-the-art works across three different tasks. Our code is available at https://github.com/hzzhou01/Closed-Boundary-Learning.", }

The Universum class, often known as the *other* class or the*miscellaneous* class, is defined as a collection of samples that do not belong to any class of interest. It is a typical class that exists in many classification-based tasks in NLP, such as relation extraction, named entity recognition, sentiment analysis, etc. The Universum class exhibits very different properties, namely heterogeneity and lack of representativeness in training data; however, existing methods often treat the Universum class equally with the classes of interest, leading to problems such as overfitting, misclassification, and diminished model robustness. In this work, we propose a closed boundary learning method that applies closed decision boundaries to classes of interest and designates the area outside all closed boundaries in the feature space as the space of the Universum class. Specifically, we formulate closed boundaries as arbitrary shapes, propose the inter-class rule-based probability estimation for the Universum class to cater to its unique properties, and propose a boundary learning loss to adjust decision boundaries based on the balance of misclassified samples inside and outside the boundary. In adherence to the natural properties of the Universum class, our method enhances both accuracy and robustness of classification models, demonstrated by improvements on six state-of-the-art works across three different tasks. Our code is available at https://github.com/hzzhou01/Closed-Boundary-Learning.

Closed Boundary Learning for Classification Tasks with the Universum Class

findings-emnlp.1038

Poster

https://aclanthology.org/2023.findings-emnlp.1039.bib

https://aclanthology.org/2023.findings-emnlp.1039/

@inproceedings{verma-etal-2023-revisiting, title = "Revisiting Entropy Rate Constancy in Text", author = "Verma, Vivek and Tomlin, Nicholas and Klein, Dan", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1039", doi = "10.18653/v1/2023.findings-emnlp.1039", pages = "15537--15549", abstract = "The uniform information density (UID) hypothesis states that humans tend to distribute information roughly evenly across an utterance or discourse. Early evidence in support of the UID hypothesis came from Genzel and Charniak (2002), which proposed an entropy rate constancy principle based on the probability of English text under $n$-gram language models. We re-evaluate the claims of Genzel and Charniak (2002) with neural language models, failing to find clear evidence in support of entropy rate constancy. We conduct a range of experiments across datasets, model sizes, and languages and discuss implications for the uniform information density hypothesis and linguistic theories of efficient communication more broadly.", }

The uniform information density (UID) hypothesis states that humans tend to distribute information roughly evenly across an utterance or discourse. Early evidence in support of the UID hypothesis came from Genzel and Charniak (2002), which proposed an entropy rate constancy principle based on the probability of English text under $n$-gram language models. We re-evaluate the claims of Genzel and Charniak (2002) with neural language models, failing to find clear evidence in support of entropy rate constancy. We conduct a range of experiments across datasets, model sizes, and languages and discuss implications for the uniform information density hypothesis and linguistic theories of efficient communication more broadly.

Revisiting Entropy Rate Constancy in Text

findings-emnlp.1039

2305.12084

https://huggingface.co/papers/2305.12084

Poster

https://aclanthology.org/2023.findings-emnlp.1040.bib

https://aclanthology.org/2023.findings-emnlp.1040/

@inproceedings{feng-etal-2023-calibrated, title = "Calibrated Seq2seq Models for Efficient and Generalizable Ultra-fine Entity Typing", author = "Feng, Yanlin and Pratapa, Adithya and Mortensen, David", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1040", doi = "10.18653/v1/2023.findings-emnlp.1040", pages = "15550--15560", abstract = "Ultra-fine entity typing plays a crucial role in information extraction by predicting fine-grained semantic types for entity mentions in text. However, this task poses significant challenges due to the massive number of entity types in the output space. The current state-of-the-art approaches, based on standard multi-label classifiers or cross-encoder models, suffer from poor generalization performance or inefficient inference speed. In this paper, we present CASENT, a seq2seq model designed for ultra-fine entity typing that predicts ultra-fine types with calibrated confidence scores. Our model takes an entity mention as input and employs constrained beam search to generate multiple types autoregressively. The raw sequence probabilities associated with the predicted types are then transformed into confidence scores using a novel calibration method. We conduct extensive experiments on the UFET dataset which contains over $10k$ types. Our method outperforms the previous state-of-the-art in terms of F1 score and calibration error, while achieving an inference speedup of over 50 times. Additionally, we demonstrate the generalization capabilities of our model by evaluating it in zero-shot and few-shot settings on five specialized domain entity typing datasets that are unseen during training. Remarkably, our model outperforms large language models with 10 times more parameters in the zero-shot setting, and when fine-tuned on 50 examples, it significantly outperforms ChatGPT on all datasets.", }

Ultra-fine entity typing plays a crucial role in information extraction by predicting fine-grained semantic types for entity mentions in text. However, this task poses significant challenges due to the massive number of entity types in the output space. The current state-of-the-art approaches, based on standard multi-label classifiers or cross-encoder models, suffer from poor generalization performance or inefficient inference speed. In this paper, we present CASENT, a seq2seq model designed for ultra-fine entity typing that predicts ultra-fine types with calibrated confidence scores. Our model takes an entity mention as input and employs constrained beam search to generate multiple types autoregressively. The raw sequence probabilities associated with the predicted types are then transformed into confidence scores using a novel calibration method. We conduct extensive experiments on the UFET dataset which contains over $10k$ types. Our method outperforms the previous state-of-the-art in terms of F1 score and calibration error, while achieving an inference speedup of over 50 times. Additionally, we demonstrate the generalization capabilities of our model by evaluating it in zero-shot and few-shot settings on five specialized domain entity typing datasets that are unseen during training. Remarkably, our model outperforms large language models with 10 times more parameters in the zero-shot setting, and when fine-tuned on 50 examples, it significantly outperforms ChatGPT on all datasets.

Calibrated Seq2seq Models for Efficient and Generalizable Ultra-fine Entity Typing

findings-emnlp.1040

2311.00835

Poster

https://aclanthology.org/2023.findings-emnlp.1041.bib

https://aclanthology.org/2023.findings-emnlp.1041/

@inproceedings{li-etal-2023-learning-semantic, title = "Learning Semantic Role Labeling from Compatible Label Sequences", author = "Li, Tao and Kazeminejad, Ghazaleh and Brown, Susan and Srikumar, Vivek and Palmer, Martha", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1041", doi = "10.18653/v1/2023.findings-emnlp.1041", pages = "15561--15572", abstract = "Semantic role labeling (SRL) has multiple disjoint label sets, e.g., VerbNet and PropBank. Creating these datasets is challenging, therefore a natural question is how to use each one to help the other. Prior work has shown that cross-task interaction helps, but only explored multitask learning so far. A common issue with multi-task setup is that argument sequences are still separately decoded, running the risk of generating structurally inconsistent label sequences (as per lexicons like Semlink). In this paper, we eliminate such issue with a framework that jointly models VerbNet and PropBank labels as one sequence. In this setup, we show that enforcing Semlink constraints during decoding constantly improves the overall F1. With special input constructions, our joint model infers VerbNet arguments from given PropBank arguments with over 99 F1. For learning, we propose a constrained marginal model that learns with knowledge defined in Semlink to further benefit from the large amounts of PropBank-only data. On the joint benchmark based on CoNLL05, our models achieve state-of-the-art F1{'}s, outperforming the prior best in-domain model by 3.5 (VerbNet) and 0.8 (PropBank). For out-of-domain generalization, our models surpass the prior best by 3.4 (VerbNet) and 0.2 (PropBank).", }

Semantic role labeling (SRL) has multiple disjoint label sets, e.g., VerbNet and PropBank. Creating these datasets is challenging, therefore a natural question is how to use each one to help the other. Prior work has shown that cross-task interaction helps, but only explored multitask learning so far. A common issue with multi-task setup is that argument sequences are still separately decoded, running the risk of generating structurally inconsistent label sequences (as per lexicons like Semlink). In this paper, we eliminate such issue with a framework that jointly models VerbNet and PropBank labels as one sequence. In this setup, we show that enforcing Semlink constraints during decoding constantly improves the overall F1. With special input constructions, our joint model infers VerbNet arguments from given PropBank arguments with over 99 F1. For learning, we propose a constrained marginal model that learns with knowledge defined in Semlink to further benefit from the large amounts of PropBank-only data. On the joint benchmark based on CoNLL05, our models achieve state-of-the-art F1{'}s, outperforming the prior best in-domain model by 3.5 (VerbNet) and 0.8 (PropBank). For out-of-domain generalization, our models surpass the prior best by 3.4 (VerbNet) and 0.2 (PropBank).

Learning Semantic Role Labeling from Compatible Label Sequences

findings-emnlp.1041

2305.14600

Poster

https://aclanthology.org/2023.findings-emnlp.1042.bib

https://aclanthology.org/2023.findings-emnlp.1042/

@inproceedings{campese-etal-2023-quadro, title = "{QUADR}o: Dataset and Models for {QU}estion-Answer Database Retrieval", author = "Campese, Stefano and Lauriola, Ivano and Moschitti, Alessandro", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1042", doi = "10.18653/v1/2023.findings-emnlp.1042", pages = "15573--15587", abstract = "An effective approach to design automated Question Answering (QA) systems is to efficiently retrieve answers from pre-computed databases containing question/answer pairs. One of the main challenges to this design is the lack of training/testing data. Existing resources are limited in size and topics and either do not consider answers (question-question similarity only) or their quality in the annotation process. To fill this gap, we introduce a novel open-domain annotated resource to train and evaluate models for this task. The resource consists of 15,211 input questions. Each question is paired with 30 similar question/answer pairs, resulting in a total of 443,000 annotated examples. The binary label associated with each pair indicates the relevance with respect to the input question. Furthermore, we report extensive experimentation to test the quality and properties of our resource with respect to various key aspects of QA systems, including answer relevance, training strategies, and models input configuration.", }

An effective approach to design automated Question Answering (QA) systems is to efficiently retrieve answers from pre-computed databases containing question/answer pairs. One of the main challenges to this design is the lack of training/testing data. Existing resources are limited in size and topics and either do not consider answers (question-question similarity only) or their quality in the annotation process. To fill this gap, we introduce a novel open-domain annotated resource to train and evaluate models for this task. The resource consists of 15,211 input questions. Each question is paired with 30 similar question/answer pairs, resulting in a total of 443,000 annotated examples. The binary label associated with each pair indicates the relevance with respect to the input question. Furthermore, we report extensive experimentation to test the quality and properties of our resource with respect to various key aspects of QA systems, including answer relevance, training strategies, and models input configuration.

QUADRo: Dataset and Models for QUestion-Answer Database Retrieval

findings-emnlp.1042

2304.01003

Poster

https://aclanthology.org/2023.findings-emnlp.1043.bib

https://aclanthology.org/2023.findings-emnlp.1043/

@inproceedings{youssef-etal-2023-give, title = "Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models", author = {Youssef, Paul and Kora{\c{s}}, Osman and Li, Meijie and Schl{\"o}tterer, J{\"o}rg and Seifert, Christin}, editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1043", doi = "10.18653/v1/2023.findings-emnlp.1043", pages = "15588--15605", abstract = "Pre-trained Language Models (PLMs) are trained on vast unlabeled data, rich in world knowledge. This fact has sparked the interest of the community in quantifying the amount of factual knowledge present in PLMs, as this explains their performance on downstream tasks, and potentially justifies their use as knowledge bases. In this work, we survey methods and datasets that are used to probe PLMs for factual knowledge. Our contributions are: (1) We propose a categorization scheme for factual probing methods that is based on how their inputs, outputs and the probed PLMs are adapted; (2) We provide an overview of the datasets used for factual probing; (3) We synthesize insights about knowledge retention and prompt optimization in PLMs, analyze obstacles to adopting PLMs as knowledge bases and outline directions for future work.", }

Pre-trained Language Models (PLMs) are trained on vast unlabeled data, rich in world knowledge. This fact has sparked the interest of the community in quantifying the amount of factual knowledge present in PLMs, as this explains their performance on downstream tasks, and potentially justifies their use as knowledge bases. In this work, we survey methods and datasets that are used to probe PLMs for factual knowledge. Our contributions are: (1) We propose a categorization scheme for factual probing methods that is based on how their inputs, outputs and the probed PLMs are adapted; (2) We provide an overview of the datasets used for factual probing; (3) We synthesize insights about knowledge retention and prompt optimization in PLMs, analyze obstacles to adopting PLMs as knowledge bases and outline directions for future work.

Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models

findings-emnlp.1043

2310.16570

https://huggingface.co/papers/2310.16570

Poster

https://aclanthology.org/2023.findings-emnlp.1044.bib

https://aclanthology.org/2023.findings-emnlp.1044/

@inproceedings{piedboeuf-langlais-2023-chatgpt, title = "Is {C}hat{GPT} the ultimate Data Augmentation Algorithm?", author = "Piedboeuf, Fr{\'e}d{\'e}ric and Langlais, Philippe", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1044", doi = "10.18653/v1/2023.findings-emnlp.1044", pages = "15606--15615", abstract = "In the aftermath of GPT-3.5, commonly known as ChatGPT, research have attempted to assess its capacity for lowering annotation cost, either by doing zero-shot learning, generating new data, or replacing human annotators. Some studies have also investigated its use for data augmentation (DA), but only in limited contexts, which still leaves the question of how ChatGPT performs compared to state-of-the-art algorithms. In this paper, we use ChatGPT to create new data both with paraphrasing and with zero-shot generation, and compare it to seven other algorithms. We show that while ChatGPT performs exceptionally well on some simpler data, it overall does not perform better than the other algorithms, yet demands a much larger implication from the practitioner due to the ChatGPT often refusing to answer due to sensitive content in the datasets.", }

In the aftermath of GPT-3.5, commonly known as ChatGPT, research have attempted to assess its capacity for lowering annotation cost, either by doing zero-shot learning, generating new data, or replacing human annotators. Some studies have also investigated its use for data augmentation (DA), but only in limited contexts, which still leaves the question of how ChatGPT performs compared to state-of-the-art algorithms. In this paper, we use ChatGPT to create new data both with paraphrasing and with zero-shot generation, and compare it to seven other algorithms. We show that while ChatGPT performs exceptionally well on some simpler data, it overall does not perform better than the other algorithms, yet demands a much larger implication from the practitioner due to the ChatGPT often refusing to answer due to sensitive content in the datasets.

Is ChatGPT the ultimate Data Augmentation Algorithm?

findings-emnlp.1044

Poster

https://aclanthology.org/2023.findings-emnlp.1045.bib

https://aclanthology.org/2023.findings-emnlp.1045/

@inproceedings{kim-cho-2023-enhanced, title = "Enhanced Simultaneous Machine Translation with Word-level Policies", author = "Kim, Kang and Cho, Hankyu", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1045", doi = "10.18653/v1/2023.findings-emnlp.1045", pages = "15616--15634", abstract = "Recent years have seen remarkable advances in the field of Simultaneous Machine Translation (SiMT) due to the introduction of innovative policies that dictate whether to READ or WRITE at each step of the translation process. However, a common assumption in many existing studies is that operations are carried out at the subword level, even though the standard unit for input and output in most practical scenarios is typically at the word level. This paper demonstrates that policies devised and validated at the subword level are surpassed by those operating at the word level, which process multiple subwords to form a complete word in a single step. Additionally, we suggest a method to boost SiMT models using language models (LMs), wherein the proposed word-level policy plays a vital role in addressing the subword disparity between LMs and SiMT models. Code is available at https://github.com/xl8-ai/WordSiMT.", }

Recent years have seen remarkable advances in the field of Simultaneous Machine Translation (SiMT) due to the introduction of innovative policies that dictate whether to READ or WRITE at each step of the translation process. However, a common assumption in many existing studies is that operations are carried out at the subword level, even though the standard unit for input and output in most practical scenarios is typically at the word level. This paper demonstrates that policies devised and validated at the subword level are surpassed by those operating at the word level, which process multiple subwords to form a complete word in a single step. Additionally, we suggest a method to boost SiMT models using language models (LMs), wherein the proposed word-level policy plays a vital role in addressing the subword disparity between LMs and SiMT models. Code is available at https://github.com/xl8-ai/WordSiMT.

Enhanced Simultaneous Machine Translation with Word-level Policies

findings-emnlp.1045

2310.16417

Poster

https://aclanthology.org/2023.findings-emnlp.1046.bib

https://aclanthology.org/2023.findings-emnlp.1046/

@inproceedings{yang-etal-2023-causal, title = "Causal Intervention-based Few-Shot Named Entity Recognition", author = "Yang, Zhen and Liu, Yongbin and Ouyang, Chunping", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1046", doi = "10.18653/v1/2023.findings-emnlp.1046", pages = "15635--15646", abstract = "Few-shot named entity recognition (NER) systems aim to recognize new classes of entities with limited labeled samples. However, these systems face a significant challenge of overfitting compared to tasks with abundant samples. This overfitting is mainly caused by the spurious correlation resulting from the bias in selecting a few samples. To address this issue, we propose a causal intervention-based few-shot NER method in this paper. Our method, based on the prototypical network, intervenes in the context to block the backdoor path between context and label. In the one-shot scenario, where no additional context is available for intervention, we employ incremental learning to intervene on the prototype, which also helps mitigate catastrophic forgetting. Our experiments on various benchmarks demonstrate that our approach achieves new state-of-the-art results.", }

Few-shot named entity recognition (NER) systems aim to recognize new classes of entities with limited labeled samples. However, these systems face a significant challenge of overfitting compared to tasks with abundant samples. This overfitting is mainly caused by the spurious correlation resulting from the bias in selecting a few samples. To address this issue, we propose a causal intervention-based few-shot NER method in this paper. Our method, based on the prototypical network, intervenes in the context to block the backdoor path between context and label. In the one-shot scenario, where no additional context is available for intervention, we employ incremental learning to intervene on the prototype, which also helps mitigate catastrophic forgetting. Our experiments on various benchmarks demonstrate that our approach achieves new state-of-the-art results.

Causal Intervention-based Few-Shot Named Entity Recognition

findings-emnlp.1046

Poster

https://aclanthology.org/2023.findings-emnlp.1047.bib

https://aclanthology.org/2023.findings-emnlp.1047/

@inproceedings{jiang-2023-tadi, title = "{TADI}: Topic-aware Attention and Powerful Dual-encoder Interaction for Recall in News Recommendation", author = "Jiang, Junxiang", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1047", doi = "10.18653/v1/2023.findings-emnlp.1047", pages = "15647--15658", abstract = "News recommendation is one of the widest commercialization in natural language processing research area, which aims to recommend news according to user interests. New recall plays an important role in news recommendation. It is to recall candidates from a very large news database. Recent researches of news recall mostly adopt dual-encoder architecture as it provides a much faster recall scheme, and they encode each word equally. However, these works remain two challenges: irrelevant word distraction and weak dual-encoder interaction. Therefore, we propose a model Topic-aware Attention and powerful Dual-encoder Interaction for Recall in news recommendation (TADI). To avoid irrelevant word distraction, TADI designs a Topic-aware Attention (TA) which weights words according to news topics. To enhance dual-encoder interaction, TADI provides a cheap yet powerful interaction module, namely Dual-encoder Interaction (DI). DI helps dual encoders interact powerfully based on two aux targets. After performance comparisons between TADI and state-of-the-arts in a series of experiments, we verify the effectiveness of TADI.", }

News recommendation is one of the widest commercialization in natural language processing research area, which aims to recommend news according to user interests. New recall plays an important role in news recommendation. It is to recall candidates from a very large news database. Recent researches of news recall mostly adopt dual-encoder architecture as it provides a much faster recall scheme, and they encode each word equally. However, these works remain two challenges: irrelevant word distraction and weak dual-encoder interaction. Therefore, we propose a model Topic-aware Attention and powerful Dual-encoder Interaction for Recall in news recommendation (TADI). To avoid irrelevant word distraction, TADI designs a Topic-aware Attention (TA) which weights words according to news topics. To enhance dual-encoder interaction, TADI provides a cheap yet powerful interaction module, namely Dual-encoder Interaction (DI). DI helps dual encoders interact powerfully based on two aux targets. After performance comparisons between TADI and state-of-the-arts in a series of experiments, we verify the effectiveness of TADI.

TADI: Topic-aware Attention and Powerful Dual-encoder Interaction for Recall in News Recommendation

findings-emnlp.1047

Poster

https://aclanthology.org/2023.findings-emnlp.1048.bib

https://aclanthology.org/2023.findings-emnlp.1048/

@inproceedings{mirzakhmedova-etal-2023-unveiling, title = "Unveiling the Power of Argument Arrangement in Online Persuasive Discussions", author = "Mirzakhmedova, Nailia and Kiesel, Johannes and Al-Khatib, Khalid and Stein, Benno", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1048", doi = "10.18653/v1/2023.findings-emnlp.1048", pages = "15659--15671", abstract = "Previous research on argumentation in online discussions has largely focused on examining individual comments and neglected the interactive nature of discussions. In line with previous work, we represent individual comments as sequences of semantic argumentative unit types. However, because it is intuitively necessary for dialogical argumentation to address the opposing viewpoints, we extend this model by clustering type sequences into different argument arrangement patterns and representing discussions as sequences of these patterns. These sequences of patterns are a symbolic representation of argumentation strategies that capture the overall structure of discussions. Using this novel approach, we conduct an in-depth analysis of the strategies in 34,393 discussions from the online discussion forum Change My View and show that our discussion model is effective for persuasiveness prediction, outperforming LLM-based classifiers on the same data. Our results provide valuable insights into argumentation dynamics in online discussions and, through the presented prediction procedure, are of practical importance for writing assistance and persuasive text generation systems.", }

Previous research on argumentation in online discussions has largely focused on examining individual comments and neglected the interactive nature of discussions. In line with previous work, we represent individual comments as sequences of semantic argumentative unit types. However, because it is intuitively necessary for dialogical argumentation to address the opposing viewpoints, we extend this model by clustering type sequences into different argument arrangement patterns and representing discussions as sequences of these patterns. These sequences of patterns are a symbolic representation of argumentation strategies that capture the overall structure of discussions. Using this novel approach, we conduct an in-depth analysis of the strategies in 34,393 discussions from the online discussion forum Change My View and show that our discussion model is effective for persuasiveness prediction, outperforming LLM-based classifiers on the same data. Our results provide valuable insights into argumentation dynamics in online discussions and, through the presented prediction procedure, are of practical importance for writing assistance and persuasive text generation systems.

Unveiling the Power of Argument Arrangement in Online Persuasive Discussions

findings-emnlp.1048

Poster

https://aclanthology.org/2023.findings-emnlp.1049.bib

https://aclanthology.org/2023.findings-emnlp.1049/

@inproceedings{ma-etal-2023-ffaeval, title = "{FFAE}val: Evaluating Dialogue System via Free-For-All Ranking", author = "Ma, Zeyao and Yao, Zijun and Zhang, Jing and Yu, Jifan and Zhang, Xiaohan and Li, Juanzi and Tang, Jie", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1049", doi = "10.18653/v1/2023.findings-emnlp.1049", pages = "15672--15684", abstract = "Evaluating open-domain dialogue systems is currently an open question. Automatic evaluation metrics have shown poor correlation with human assessment in dialogue generation tasks. Human evaluation, which involves annotators for multi-dimension scoring, is trustworthy but time-consuming. In this work, we propose FFAEval, a reliable and efficient human evaluation framework using Free-For-All ranking approach. By sharing the dialogue history, the framework enables annotators to converse with multiple dialogue systems simultaneously in a single-blind, multi-turn manner. The subsequent free-for-all allows annotators to select the most favourable model in each turn from among all the participating dialogue systems. The final performance of each model is represented by calculating the TrueSkill score derived from the free-for-all competition. Our empirical study on English and Chinese dialogue systems demonstrates that FFAEval achieves a strong correlation with score-based human assessment compared to existing evaluation methods. We further prove the efficiency and stability of our framework in additional experiments. The source code and data are available on Github.", }

Evaluating open-domain dialogue systems is currently an open question. Automatic evaluation metrics have shown poor correlation with human assessment in dialogue generation tasks. Human evaluation, which involves annotators for multi-dimension scoring, is trustworthy but time-consuming. In this work, we propose FFAEval, a reliable and efficient human evaluation framework using Free-For-All ranking approach. By sharing the dialogue history, the framework enables annotators to converse with multiple dialogue systems simultaneously in a single-blind, multi-turn manner. The subsequent free-for-all allows annotators to select the most favourable model in each turn from among all the participating dialogue systems. The final performance of each model is represented by calculating the TrueSkill score derived from the free-for-all competition. Our empirical study on English and Chinese dialogue systems demonstrates that FFAEval achieves a strong correlation with score-based human assessment compared to existing evaluation methods. We further prove the efficiency and stability of our framework in additional experiments. The source code and data are available on Github.

FFAEval: Evaluating Dialogue System via Free-For-All Ranking

findings-emnlp.1049

Poster

https://aclanthology.org/2023.findings-emnlp.1050.bib

https://aclanthology.org/2023.findings-emnlp.1050/

@inproceedings{chen-etal-2023-orca, title = "Orca: A Few-shot Benchmark for {C}hinese Conversational Machine Reading Comprehension", author = "Chen, Nuo and Li, Hongguang and He, Junqing and Bao, Yinan and Lin, Xinshi and Yang, Qi and Liu, Jianfeng and Gan, Ruyi and Zhang, Jiaxing and Wang, Baoyuan and Li, Jia", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1050", doi = "10.18653/v1/2023.findings-emnlp.1050", pages = "15685--15699", abstract = "The conversational machine reading comprehension (CMRC) task aims to answer questions in conversations, which has been a hot research topic in recent years because of its wide applications. However, existing CMRC benchmarks in which each conversation is assigned a static passage are inconsistent with real scenarios. Thus, model{'}s comprehension ability towards real scenarios are hard to evaluate reasonably. To this end, we propose the first Chinese CMRC benchmark \textbf{Orca} and further provide zero-shot/few-shot settings to evaluate model{'}s generalization ability towards diverse domains. We collect 831 hot-topic driven conversations with 4,742 turns in total. Each turn of a conversation is assigned with a response-related passage, aiming to evaluate model{'}s comprehension ability more reasonably. The topics of conversations are collected from social media platform and cover 33 domains, trying to be consistent with real scenarios. Importantly, answers in Orca are all well-annotated natural responses rather than the specific spans or short phrase in previous datasets. Besides, we implement three strong baselines to tackle the challenge in Orca. The results indicate the great challenge of our CMRC benchmark.", }

The conversational machine reading comprehension (CMRC) task aims to answer questions in conversations, which has been a hot research topic in recent years because of its wide applications. However, existing CMRC benchmarks in which each conversation is assigned a static passage are inconsistent with real scenarios. Thus, model{'}s comprehension ability towards real scenarios are hard to evaluate reasonably. To this end, we propose the first Chinese CMRC benchmark \textbf{Orca} and further provide zero-shot/few-shot settings to evaluate model{'}s generalization ability towards diverse domains. We collect 831 hot-topic driven conversations with 4,742 turns in total. Each turn of a conversation is assigned with a response-related passage, aiming to evaluate model{'}s comprehension ability more reasonably. The topics of conversations are collected from social media platform and cover 33 domains, trying to be consistent with real scenarios. Importantly, answers in Orca are all well-annotated natural responses rather than the specific spans or short phrase in previous datasets. Besides, we implement three strong baselines to tackle the challenge in Orca. The results indicate the great challenge of our CMRC benchmark.

Orca: A Few-shot Benchmark for Chinese Conversational Machine Reading Comprehension

findings-emnlp.1050

2302.13619

Poster

https://aclanthology.org/2023.findings-emnlp.1051.bib

https://aclanthology.org/2023.findings-emnlp.1051/

@inproceedings{huang-chang-2023-ver, title = "{VER}: Unifying Verbalizing Entities and Relations", author = "Huang, Jie and Chang, Kevin", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1051", doi = "10.18653/v1/2023.findings-emnlp.1051", pages = "15700--15710", abstract = "Entities and relationships between entities are vital in the real world. Essentially, we understand the world by understanding entities and relations. For instance, to understand a field, e.g., computer science, we need to understand the relevant concepts, e.g., machine learning, and the relationships between concepts, e.g., machine learning and artificial intelligence. To understand a person, we should first know who he/she is and how he/she is related to others. To understand entities and relations, humans may refer to natural language descriptions. For instance, when learning a new scientific term, people usually start by reading its definition in dictionaries or encyclopedias. To know the relationship between two entities, humans tend to create a sentence to connect them. In this paper, we propose VER: a unified model for Verbalizing Entities and Relations. Specifically, we attempt to build a system that takes any entity or entity set as input and generates a sentence to represent entities and relations. Extensive experiments demonstrate that our model can generate high-quality sentences describing entities and entity relationships and facilitate various tasks on entities and relations, including definition modeling, relation modeling, and generative commonsense reasoning.", }

Entities and relationships between entities are vital in the real world. Essentially, we understand the world by understanding entities and relations. For instance, to understand a field, e.g., computer science, we need to understand the relevant concepts, e.g., machine learning, and the relationships between concepts, e.g., machine learning and artificial intelligence. To understand a person, we should first know who he/she is and how he/she is related to others. To understand entities and relations, humans may refer to natural language descriptions. For instance, when learning a new scientific term, people usually start by reading its definition in dictionaries or encyclopedias. To know the relationship between two entities, humans tend to create a sentence to connect them. In this paper, we propose VER: a unified model for Verbalizing Entities and Relations. Specifically, we attempt to build a system that takes any entity or entity set as input and generates a sentence to represent entities and relations. Extensive experiments demonstrate that our model can generate high-quality sentences describing entities and entity relationships and facilitate various tasks on entities and relations, including definition modeling, relation modeling, and generative commonsense reasoning.

VER: Unifying Verbalizing Entities and Relations

findings-emnlp.1051

2211.11093

Poster

https://aclanthology.org/2023.findings-emnlp.1052.bib

https://aclanthology.org/2023.findings-emnlp.1052/

@inproceedings{xu-etal-2023-linearity, title = "The Linearity of the Effect of Surprisal on Reading Times across Languages", author = "Xu, Weijie and Chon, Jason and Liu, Tianran and Futrell, Richard", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1052", doi = "10.18653/v1/2023.findings-emnlp.1052", pages = "15711--15721", abstract = "In psycholinguistics, surprisal theory posits that the amount of online processing effort expended by a human comprehender per word positively correlates with the surprisal of that word given its preceding context. In addition to this overall correlation, more importantly, the specific quantitative form taken by the processing effort as a function of surprisal offers insights into the underlying cognitive mechanisms of language processing. Focusing on English, previous studies have looked into the linearity of surprisal on reading times. Here, we extend the investigation by examining eyetracking corpora of seven languages: Danish, Dutch, English, German, Japanese, Mandarin, and Russian. We find evidence for superlinearity in some languages, but the results are highly sensitive to which language model is used to estimate surprisal.", }

In psycholinguistics, surprisal theory posits that the amount of online processing effort expended by a human comprehender per word positively correlates with the surprisal of that word given its preceding context. In addition to this overall correlation, more importantly, the specific quantitative form taken by the processing effort as a function of surprisal offers insights into the underlying cognitive mechanisms of language processing. Focusing on English, previous studies have looked into the linearity of surprisal on reading times. Here, we extend the investigation by examining eyetracking corpora of seven languages: Danish, Dutch, English, German, Japanese, Mandarin, and Russian. We find evidence for superlinearity in some languages, but the results are highly sensitive to which language model is used to estimate surprisal.

The Linearity of the Effect of Surprisal on Reading Times across Languages

findings-emnlp.1052

Poster

https://aclanthology.org/2023.findings-emnlp.1053.bib

https://aclanthology.org/2023.findings-emnlp.1053/

@inproceedings{li-etal-2023-adversarial, title = "Adversarial Text Generation by Search and Learning", author = "Li, Guoyi and Shi, Bingkang and Liu, Zongzhen and Kong, Dehan and Wu, Yulei and Zhang, Xiaodan and Huang, Longtao and Lyu, Honglei", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1053", doi = "10.18653/v1/2023.findings-emnlp.1053", pages = "15722--15738", abstract = "Recent research has shown that evaluating the robustness of natural language processing models using textual attack methods is significant. However, most existing text attack methods only use heuristic replacement strategies or language models to generate replacement words at the word level. The blind pursuit of high attack success rates makes it difficult to ensure the quality of the generated adversarial text. As a result, adversarial text is often difficult for humans to understand. In fact, many methods that perform well in terms of text attacks often generate adversarial text with poor quality. To address this important gap, our work treats black-box text attack as an unsupervised text generation problem and proposes a search and learning framework for Adversarial Text Generation by Search and Learning (ATGSL) and develops three adversarial attack methods (ATGSL-SA, ATGSL-BM, ATGSL-FUSION) for black box text attacks. We first apply a heuristic search attack algorithm (ATGSL-SA) and a linguistic thesaurus to generate adversarial samples with high semantic similarity. After this process, we train a conditional generative model to learn from the search results while smoothing out search noise. Moreover, we design an efficient ATGSL-BM attack algorithm based on the text generator. Furthermore, we propose a hybrid attack method (ATGSL-FUSION) that integrates the advantages of ATGSL-SA and ATGSL-BM to enhance attack effectiveness. Our proposed attack algorithms are significantly superior to the most advanced methods in terms of attack efficiency and adversarial text quality.", }

Recent research has shown that evaluating the robustness of natural language processing models using textual attack methods is significant. However, most existing text attack methods only use heuristic replacement strategies or language models to generate replacement words at the word level. The blind pursuit of high attack success rates makes it difficult to ensure the quality of the generated adversarial text. As a result, adversarial text is often difficult for humans to understand. In fact, many methods that perform well in terms of text attacks often generate adversarial text with poor quality. To address this important gap, our work treats black-box text attack as an unsupervised text generation problem and proposes a search and learning framework for Adversarial Text Generation by Search and Learning (ATGSL) and develops three adversarial attack methods (ATGSL-SA, ATGSL-BM, ATGSL-FUSION) for black box text attacks. We first apply a heuristic search attack algorithm (ATGSL-SA) and a linguistic thesaurus to generate adversarial samples with high semantic similarity. After this process, we train a conditional generative model to learn from the search results while smoothing out search noise. Moreover, we design an efficient ATGSL-BM attack algorithm based on the text generator. Furthermore, we propose a hybrid attack method (ATGSL-FUSION) that integrates the advantages of ATGSL-SA and ATGSL-BM to enhance attack effectiveness. Our proposed attack algorithms are significantly superior to the most advanced methods in terms of attack efficiency and adversarial text quality.

Adversarial Text Generation by Search and Learning

findings-emnlp.1053

Poster

https://aclanthology.org/2023.findings-emnlp.1054.bib

https://aclanthology.org/2023.findings-emnlp.1054/

@inproceedings{lu-etal-2023-measuring, title = "Measuring Pointwise $\mathcal{V}$-Usable Information In-Context-ly", author = "Lu, Sheng and Chen, Shan and Li, Yingya and Bitterman, Danielle and Savova, Guergana and Gurevych, Iryna", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1054", doi = "10.18653/v1/2023.findings-emnlp.1054", pages = "15739--15756", abstract = "In-context learning (ICL) is a new learning paradigm that has gained popularity along with the development of large language models. In this work, we adapt a recently proposed hardness metric, pointwise $\mathcal{V}$-usable information (PVI), to an in-context version (in-context PVI). Compared to the original PVI, in-context PVI is more efficient in that it requires only a few exemplars and does not require fine-tuning. We conducted a comprehensive empirical analysis to evaluate the reliability of in-context PVI. Our findings indicate that in-context PVI estimates exhibit similar characteristics to the original PVI. Specific to the in-context setting, we show that in-context PVI estimates remain consistent across different exemplar selections and numbers of shots. The variance of in-context PVI estimates across different exemplar selections is insignificant, which suggests that in-context PVI estimates are stable. Furthermore, we demonstrate how in-context PVI can be employed to identify challenging instances. Our work highlights the potential of in-context PVI and provides new insights into the capabilities of ICL.", }

In-context learning (ICL) is a new learning paradigm that has gained popularity along with the development of large language models. In this work, we adapt a recently proposed hardness metric, pointwise $\mathcal{V}$-usable information (PVI), to an in-context version (in-context PVI). Compared to the original PVI, in-context PVI is more efficient in that it requires only a few exemplars and does not require fine-tuning. We conducted a comprehensive empirical analysis to evaluate the reliability of in-context PVI. Our findings indicate that in-context PVI estimates exhibit similar characteristics to the original PVI. Specific to the in-context setting, we show that in-context PVI estimates remain consistent across different exemplar selections and numbers of shots. The variance of in-context PVI estimates across different exemplar selections is insignificant, which suggests that in-context PVI estimates are stable. Furthermore, we demonstrate how in-context PVI can be employed to identify challenging instances. Our work highlights the potential of in-context PVI and provides new insights into the capabilities of ICL.

Measuring Pointwise 𝒱-Usable Information In-Context-ly

findings-emnlp.1054

Poster

https://aclanthology.org/2023.findings-emnlp.1055.bib

https://aclanthology.org/2023.findings-emnlp.1055/

@inproceedings{zhang-etal-2023-speechgpt, title = "{S}peech{GPT}: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities", author = "Zhang, Dong and Li, Shimin and Zhang, Xin and Zhan, Jun and Wang, Pengyu and Zhou, Yaqian and Qiu, Xipeng", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1055", doi = "10.18653/v1/2023.findings-emnlp.1055", pages = "15757--15773", abstract = "Multi-modal large language models are regarded as a crucial step towards Artificial General Intelligence (AGI) and have garnered significant interest with the emergence of ChatGPT. However, current speech-language models typically adopt the cascade paradigm, preventing inter-modal knowledge transfer. In this paper, we propose SpeechGPT, a large language model with intrinsic cross-modal conversational abilities, capable of perceiving and generating multi-modal content. With discrete speech representations, we construct SpeechInstruct, the first large-scale cross-modal speech instruction dataset. Additionally, we employ a three-stage training strategy that includes modality-adaptation pre-training, cross-modal instruction fine-tuning, and chain-of-modality instruction fine-tuning. The experimental results demonstrate that SpeechGPT has an impressive capacity to follow cross-modal human instructions and highlight the potential of handling multiple modalities with one model. Code and models are available in \url{https://github.com/0nutation/SpeechGPT}. Demos are shown in \url{https://0nutation.github.io/SpeechGPT.github.io/}.", }

Multi-modal large language models are regarded as a crucial step towards Artificial General Intelligence (AGI) and have garnered significant interest with the emergence of ChatGPT. However, current speech-language models typically adopt the cascade paradigm, preventing inter-modal knowledge transfer. In this paper, we propose SpeechGPT, a large language model with intrinsic cross-modal conversational abilities, capable of perceiving and generating multi-modal content. With discrete speech representations, we construct SpeechInstruct, the first large-scale cross-modal speech instruction dataset. Additionally, we employ a three-stage training strategy that includes modality-adaptation pre-training, cross-modal instruction fine-tuning, and chain-of-modality instruction fine-tuning. The experimental results demonstrate that SpeechGPT has an impressive capacity to follow cross-modal human instructions and highlight the potential of handling multiple modalities with one model. Code and models are available in \url{https://github.com/0nutation/SpeechGPT}. Demos are shown in \url{https://0nutation.github.io/SpeechGPT.github.io/}.

SpeechGPT: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities

findings-emnlp.1055

2305.11000

https://huggingface.co/papers/2305.11000

Poster

https://aclanthology.org/2023.findings-emnlp.1056.bib

https://aclanthology.org/2023.findings-emnlp.1056/

@inproceedings{nie-etal-2023-unleashing, title = "Unleashing the Multilingual Encoder Potential: Boosting Zero-Shot Performance via Probability Calibration", author = "Nie, Ercong and Schmid, Helmut and Schuetze, Hinrich", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1056", doi = "10.18653/v1/2023.findings-emnlp.1056", pages = "15774--15782", abstract = "Pretrained multilingual encoder models can directly perform zero-shot multilingual tasks or linguistic probing by reformulating the input examples into cloze-style prompts. This is accomplished by predicting the probabilities of the label words at the masked token position, without requiring any updates to the model parameters. However, the performance of this method is limited by the model{'}s bias toward predicting label words which frequently occurred during the pretraining. These words typically receive high probabilities. To address this issue, we combine the models with calibration techniques which modify the probabilities of label words predicted by the models. We first validate the effectiveness of a proposed simple calibration method together with other existing techniques on monolingual encoders in both zero- and few-shot scenarios. We subsequently employ these calibration techniques on multilingual encoders, resulting in substantial performance improvements across a wide range of tasks.", }

Pretrained multilingual encoder models can directly perform zero-shot multilingual tasks or linguistic probing by reformulating the input examples into cloze-style prompts. This is accomplished by predicting the probabilities of the label words at the masked token position, without requiring any updates to the model parameters. However, the performance of this method is limited by the model{'}s bias toward predicting label words which frequently occurred during the pretraining. These words typically receive high probabilities. To address this issue, we combine the models with calibration techniques which modify the probabilities of label words predicted by the models. We first validate the effectiveness of a proposed simple calibration method together with other existing techniques on monolingual encoders in both zero- and few-shot scenarios. We subsequently employ these calibration techniques on multilingual encoders, resulting in substantial performance improvements across a wide range of tasks.

Unleashing the Multilingual Encoder Potential: Boosting Zero-Shot Performance via Probability Calibration

findings-emnlp.1056

2310.05069

Poster

https://aclanthology.org/2023.findings-emnlp.1057.bib

https://aclanthology.org/2023.findings-emnlp.1057/

@inproceedings{ren-etal-2023-thorough, title = "A Thorough Examination on Zero-shot Dense Retrieval", author = "Ren, Ruiyang and Qu, Yingqi and Liu, Jing and Zhao, Xin and Wu, Qifei and Ding, Yuchen and Wu, Hua and Wang, Haifeng and Wen, Ji-Rong", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1057", doi = "10.18653/v1/2023.findings-emnlp.1057", pages = "15783--15796", abstract = "Recent years have witnessed the significant advance in dense retrieval (DR) based on powerful pre-trained language models (PLM). DR models have achieved excellent performance in several benchmark datasets, while they are shown to be not as competitive as traditional sparse retrieval models (e.g., BM25) in a zero-shot retrieval setting. However, in the related literature, there still lacks a detailed and comprehensive study on zero-shot retrieval. In this paper, we present the first thorough examination of the zero-shot capability of DR models. We aim to identify the key factors and analyze how they affect zero-shot retrieval performance. In particular, we discuss the effect of several key factors related to source training set, analyze the potential bias from the target dataset, and review and compare existing zero-shot DR models. Our findings provide important evidence to better understand and develop zero-shot DR models.", }

Recent years have witnessed the significant advance in dense retrieval (DR) based on powerful pre-trained language models (PLM). DR models have achieved excellent performance in several benchmark datasets, while they are shown to be not as competitive as traditional sparse retrieval models (e.g., BM25) in a zero-shot retrieval setting. However, in the related literature, there still lacks a detailed and comprehensive study on zero-shot retrieval. In this paper, we present the first thorough examination of the zero-shot capability of DR models. We aim to identify the key factors and analyze how they affect zero-shot retrieval performance. In particular, we discuss the effect of several key factors related to source training set, analyze the potential bias from the target dataset, and review and compare existing zero-shot DR models. Our findings provide important evidence to better understand and develop zero-shot DR models.

A Thorough Examination on Zero-shot Dense Retrieval

findings-emnlp.1057

2204.12755

Poster

https://aclanthology.org/2023.findings-emnlp.1058.bib

https://aclanthology.org/2023.findings-emnlp.1058/

@inproceedings{peng-etal-2023-contrastive, title = "Contrastive Pre-training for Personalized Expert Finding", author = "Peng, Qiyao and Liu, Hongtao and Lv, Zhepeng and Yang, Qing and Wang, Wenjun", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1058", doi = "10.18653/v1/2023.findings-emnlp.1058", pages = "15797--15806", abstract = "Expert finding could help route questions to potential suitable users to answer in Community Question Answering (CQA) platforms. Hence it is essential to learn accurate representations of experts and questions according to the question text articles. Recently the pre-training and fine-tuning paradigms are powerful for natural language understanding, which has the potential for better question modeling and expert finding. Inspired by this, we propose a CQA-domain Contrastive Pre-training framework for Expert Finding, named CPEF, which could learn more comprehensive question representations. Specifically, considering that there is semantic complementation between question titles and bodies, during the domain pre-training phase, we propose a title-body contrastive learning task to enhance question representations, which directly treats the question title and the corresponding body as positive samples of each other, instead of designing extra data-augmentation strategies. Furthermore, a personalized tuning network is proposed to inject the personalized preferences of different experts during the fine-tuning phase. Extensive experimental results on six real-world datasets demonstrate that our method could achieve superior performance for expert finding.", }

Expert finding could help route questions to potential suitable users to answer in Community Question Answering (CQA) platforms. Hence it is essential to learn accurate representations of experts and questions according to the question text articles. Recently the pre-training and fine-tuning paradigms are powerful for natural language understanding, which has the potential for better question modeling and expert finding. Inspired by this, we propose a CQA-domain Contrastive Pre-training framework for Expert Finding, named CPEF, which could learn more comprehensive question representations. Specifically, considering that there is semantic complementation between question titles and bodies, during the domain pre-training phase, we propose a title-body contrastive learning task to enhance question representations, which directly treats the question title and the corresponding body as positive samples of each other, instead of designing extra data-augmentation strategies. Furthermore, a personalized tuning network is proposed to inject the personalized preferences of different experts during the fine-tuning phase. Extensive experimental results on six real-world datasets demonstrate that our method could achieve superior performance for expert finding.

Contrastive Pre-training for Personalized Expert Finding

findings-emnlp.1058

Poster

https://aclanthology.org/2023.findings-emnlp.1059.bib

https://aclanthology.org/2023.findings-emnlp.1059/

@inproceedings{shen-etal-2023-mitigating, title = "Mitigating Intrinsic Named Entity-Related Hallucinations of Abstractive Text Summarization", author = "Shen, Jianbin and Xuan, Junyu and Liang, Christy", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1059", doi = "10.18653/v1/2023.findings-emnlp.1059", pages = "15807--15824", abstract = "Abstractive text summarization (ATS) is both important and challenging. Recent studies have shown that ATS still faces various forms of hallucination. Our study also indicates that a significant portion of hallucinations is named entity-related. They might appear in different forms, such as mistaken entities and erroneous entity references. The underlying causes implicit in data are complex: data samples pose varying learning conditions. Despite recent research efforts dedicated to named entity-related hallucinations, the solutions have not adequately addressed the varying learning conditions posed by data. This paper aims to bridge the gap in pursuit of reducing intrinsic named entity-related hallucinations. To do so, we propose an adaptive margin ranking loss to facilitate two entity-alignment learning methods to tackle them. Our experiment results show that our methods improve the used baseline model on automatic evaluation scores. The human evaluation also indicates that our methods jointly reduce the intrinsic named entity-related hallucinations considerably compared to the used baseline model.", }

Abstractive text summarization (ATS) is both important and challenging. Recent studies have shown that ATS still faces various forms of hallucination. Our study also indicates that a significant portion of hallucinations is named entity-related. They might appear in different forms, such as mistaken entities and erroneous entity references. The underlying causes implicit in data are complex: data samples pose varying learning conditions. Despite recent research efforts dedicated to named entity-related hallucinations, the solutions have not adequately addressed the varying learning conditions posed by data. This paper aims to bridge the gap in pursuit of reducing intrinsic named entity-related hallucinations. To do so, we propose an adaptive margin ranking loss to facilitate two entity-alignment learning methods to tackle them. Our experiment results show that our methods improve the used baseline model on automatic evaluation scores. The human evaluation also indicates that our methods jointly reduce the intrinsic named entity-related hallucinations considerably compared to the used baseline model.

Mitigating Intrinsic Named Entity-Related Hallucinations of Abstractive Text Summarization

findings-emnlp.1059

Poster

https://aclanthology.org/2023.findings-emnlp.1060.bib

https://aclanthology.org/2023.findings-emnlp.1060/

@inproceedings{liu-wang-2023-towards, title = "Towards Informative Few-Shot Prompt with Maximum Information Gain for In-Context Learning", author = "Liu, Hongfu and Wang, Ye", editor = "Bouamor, Houda and Pino, Juan and Bali, Kalika", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023", month = dec, year = "2023", address = "Singapore", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-emnlp.1060", doi = "10.18653/v1/2023.findings-emnlp.1060", pages = "15825--15838", abstract = "Large Language models (LLMs) possess the capability to engage In-context Learning (ICL) by leveraging a few demonstrations pertaining to a new downstream task as conditions. However, this particular learning paradigm suffers from high instability stemming from substantial variances induced by factors such as the input distribution of selected examples, their ordering, and prompt formats. In this work, we demonstrate that even when all these factors are held constant, the random selection of examples still results in high variance. Consequently, we aim to explore the informative ability of data examples by quantifying the Information Gain (IG) obtained in prediction after observing a given example candidate. Then we propose to sample those with maximum IG. Additionally, we identify the presence of template bias, which can lead to unfair evaluations of IG during the sampling process. To mitigate this bias, we introduce Calibration Before Sampling strategy. The experimental results illustrate that our proposed method can yield an average relative improvement of 14.3{\%} across six classification tasks using three LLMs.", }

Large Language models (LLMs) possess the capability to engage In-context Learning (ICL) by leveraging a few demonstrations pertaining to a new downstream task as conditions. However, this particular learning paradigm suffers from high instability stemming from substantial variances induced by factors such as the input distribution of selected examples, their ordering, and prompt formats. In this work, we demonstrate that even when all these factors are held constant, the random selection of examples still results in high variance. Consequently, we aim to explore the informative ability of data examples by quantifying the Information Gain (IG) obtained in prediction after observing a given example candidate. Then we propose to sample those with maximum IG. Additionally, we identify the presence of template bias, which can lead to unfair evaluations of IG during the sampling process. To mitigate this bias, we introduce Calibration Before Sampling strategy. The experimental results illustrate that our proposed method can yield an average relative improvement of 14.3{\%} across six classification tasks using three LLMs.

Towards Informative Few-Shot Prompt with Maximum Information Gain for In-Context Learning

findings-emnlp.1060

2310.08923

Poster

https://aclanthology.org/2023.arabicnlp-1.1.bib

https://aclanthology.org/2023.arabicnlp-1.1/

@inproceedings{mohamed-etal-2023-violet, title = "Violet: A Vision-Language Model for {A}rabic Image Captioning with Gemini Decoder", author = "Mohamed, Abdelrahman and Alwajih, Fakhraddin and Nagoudi, El Moatez Billah and Inciarte, Alcides and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.1", doi = "10.18653/v1/2023.arabicnlp-1.1", pages = "1--11", abstract = "Although image captioning has a vast array of applications, it has not reached its full potential in languages other than English. Arabic, for instance, although the native language of more than 400 million people, remains largely underrepresented in this area. This is due to the lack of labeled data and powerful Arabic generative models. We alleviate this issue by presenting a novel vision-language model dedicated to Arabic, dubbed Violet. Our model is based on a vision encoder and a Gemini text decoder that maintains generation fluency while allowing fusion between the vision and language components. To train our model, we introduce a new method for automatically acquiring data from available English datasets. We also manually prepare a new dataset for evaluation. Violet performs sizeably better than our baselines on all of our evaluation datasets. For example, it reaches a CIDEr score of 61.2 on our manually annotated dataset and achieves an improvement of 13 points on Flickr8k.", }

Although image captioning has a vast array of applications, it has not reached its full potential in languages other than English. Arabic, for instance, although the native language of more than 400 million people, remains largely underrepresented in this area. This is due to the lack of labeled data and powerful Arabic generative models. We alleviate this issue by presenting a novel vision-language model dedicated to Arabic, dubbed Violet. Our model is based on a vision encoder and a Gemini text decoder that maintains generation fluency while allowing fusion between the vision and language components. To train our model, we introduce a new method for automatically acquiring data from available English datasets. We also manually prepare a new dataset for evaluation. Violet performs sizeably better than our baselines on all of our evaluation datasets. For example, it reaches a CIDEr score of 61.2 on our manually annotated dataset and achieves an improvement of 13 points on Flickr8k.

Violet: A Vision-Language Model for Arabic Image Captioning with Gemini Decoder

arabicnlp-1.1

2311.08844

Poster

https://aclanthology.org/2023.arabicnlp-1.2.bib

https://aclanthology.org/2023.arabicnlp-1.2/

@inproceedings{nayouf-etal-2023-nabra, title = "N{\^a}bra: {S}yrian {A}rabic Dialects with Morphological Annotations", author = "Nayouf, Amal and Hammouda, Tymaa and Jarrar, Mustafa and Zaraket, Fadi and Kurdy, Mohamad-Bassam", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.2", doi = "10.18653/v1/2023.arabicnlp-1.2", pages = "12--23", abstract = "This paper presents N{\^a}bra (نَبْرَة), a corpora of Syrian Arabic dialects with morphological annotations. A team of Syrian natives collected more than $6K$ sentences containing about $60K$ words from several sources including social media posts, scripts of movies and series, lyrics of songs and local proverbs to build N{\^a}bra. N{\^a}bra covers several local Syrian dialects including those of Aleppo, Damascus, Deir-ezzur, Hama, Homs, Huran, Latakia, Mardin, Raqqah, and Suwayda. A team of nine annotators annotated the $60K$ tokens with full morphological annotations across sentence contexts. We trained the annotators to follow methodological annotation guidelines to ensure unique morpheme annotations, and normalized the annotations. F1 and $\kappa$ agreement scores ranged between 74{\%} and 98{\%} across features, showing the excellent quality of N{\^a}bra annotations. Our corpora are open-source and publicly available as part of the Currasat portal https://sina.birzeit.edu/currasat.", }

This paper presents N{\^a}bra (نَبْرَة), a corpora of Syrian Arabic dialects with morphological annotations. A team of Syrian natives collected more than $6K$ sentences containing about $60K$ words from several sources including social media posts, scripts of movies and series, lyrics of songs and local proverbs to build N{\^a}bra. N{\^a}bra covers several local Syrian dialects including those of Aleppo, Damascus, Deir-ezzur, Hama, Homs, Huran, Latakia, Mardin, Raqqah, and Suwayda. A team of nine annotators annotated the $60K$ tokens with full morphological annotations across sentence contexts. We trained the annotators to follow methodological annotation guidelines to ensure unique morpheme annotations, and normalized the annotations. F1 and $\kappa$ agreement scores ranged between 74{\%} and 98{\%} across features, showing the excellent quality of N{\^a}bra annotations. Our corpora are open-source and publicly available as part of the Currasat portal https://sina.birzeit.edu/currasat.

Nâbra: Syrian Arabic Dialects with Morphological Annotations

arabicnlp-1.2

Poster

https://aclanthology.org/2023.arabicnlp-1.3.bib

https://aclanthology.org/2023.arabicnlp-1.3/

@inproceedings{ismail-etal-2023-hicma, title = "{HICMA}: The Handwriting Identification for Calligraphy and Manuscripts in {A}rabic Dataset", author = "Ismail, Anis and Kamel, Zena and Mahmoud, Reem", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.3", doi = "10.18653/v1/2023.arabicnlp-1.3", pages = "24--32", abstract = "Arabic is one of the most globally spoken languages with more than 313 million speakers worldwide. Arabic handwriting is known for its cursive nature and the variety of writing styles used. Despite the increase in effort to digitize artistic and historical elements, no public dataset was released to deal with Arabic text recognition for realistic manuscripts and calligraphic text. We present the Handwriting Identification of Manuscripts and Calligraphy in Arabic (HICMA) dataset as the first publicly available dataset with real-world and diverse samples of Arabic handwritten text in manuscripts and calligraphy. With more than 5,000 images across five different styles, the HICMA dataset includes image-text pairs and style labels for all images. We further present a comparison of the current state-of-the-art optical character recognition models in Arabic and benchmark their performance on the HICMA dataset, which serves as a baseline for future works. Both the HICMA dataset and its benchmarking tool are made available to the public under the CC BY-NC 4.0 license in the hope that the presented work opens the door to further enhancements of complex Arabic text recognition.", }

Arabic is one of the most globally spoken languages with more than 313 million speakers worldwide. Arabic handwriting is known for its cursive nature and the variety of writing styles used. Despite the increase in effort to digitize artistic and historical elements, no public dataset was released to deal with Arabic text recognition for realistic manuscripts and calligraphic text. We present the Handwriting Identification of Manuscripts and Calligraphy in Arabic (HICMA) dataset as the first publicly available dataset with real-world and diverse samples of Arabic handwritten text in manuscripts and calligraphy. With more than 5,000 images across five different styles, the HICMA dataset includes image-text pairs and style labels for all images. We further present a comparison of the current state-of-the-art optical character recognition models in Arabic and benchmark their performance on the HICMA dataset, which serves as a baseline for future works. Both the HICMA dataset and its benchmarking tool are made available to the public under the CC BY-NC 4.0 license in the hope that the presented work opens the door to further enhancements of complex Arabic text recognition.

HICMA: The Handwriting Identification for Calligraphy and Manuscripts in Arabic Dataset

arabicnlp-1.3

Poster

https://aclanthology.org/2023.arabicnlp-1.4.bib

https://aclanthology.org/2023.arabicnlp-1.4/

@inproceedings{kocaman-etal-2023-automated, title = "Automated De-Identification of {A}rabic Medical Records", author = "Kocaman, Veysel and Mellah, Youssef and Haq, Hasham and Talby, David", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.4", doi = "10.18653/v1/2023.arabicnlp-1.4", pages = "33--40", abstract = "As Electronic Health Records (EHR) become ubiquitous in healthcare systems worldwide, including in Arabic-speaking countries, the dual imperative of safeguarding patient privacy and leveraging data for research and quality improvement grows. This paper presents a first-of-its-kind automated de-identification pipeline for medical text specifically tailored for the Arabic language. This includes accurate medical Named Entity Recognition (NER) for identifying personal information; data obfuscation models to replace sensitive entities with fake entities; and an implementation that natively scales to large datasets on commodity clusters. This research makes two contributions. First, we adapt two existing NER architectures{---} BERT For Token Classification (BFTC) and BiLSTM-CNN-Char {--} to accommodate the unique syntactic and morphological characteristics of the Arabic language. Comparative analysis suggests that BFTC models outperform Bi-LSTM models, achieving higher F1 scores for both identifying and redacting personally identifiable information (PII) from Arabic medical texts. Second, we augment the deep learning models with a contextual parser engine to handle commonly missed entities. Experiments show that the combined pipeline demonstrates superior performance with micro F1 scores ranging from 0.94 to 0.98 on the test dataset, which is a translated version of the i2b2 2014 de-identification challenge, across 17 sensitive entities. This level of accuracy is in line with that achieved with manual de-identification by domain experts, suggesting that a fully automated and scalable process is now viable.", }

As Electronic Health Records (EHR) become ubiquitous in healthcare systems worldwide, including in Arabic-speaking countries, the dual imperative of safeguarding patient privacy and leveraging data for research and quality improvement grows. This paper presents a first-of-its-kind automated de-identification pipeline for medical text specifically tailored for the Arabic language. This includes accurate medical Named Entity Recognition (NER) for identifying personal information; data obfuscation models to replace sensitive entities with fake entities; and an implementation that natively scales to large datasets on commodity clusters. This research makes two contributions. First, we adapt two existing NER architectures{---} BERT For Token Classification (BFTC) and BiLSTM-CNN-Char {--} to accommodate the unique syntactic and morphological characteristics of the Arabic language. Comparative analysis suggests that BFTC models outperform Bi-LSTM models, achieving higher F1 scores for both identifying and redacting personally identifiable information (PII) from Arabic medical texts. Second, we augment the deep learning models with a contextual parser engine to handle commonly missed entities. Experiments show that the combined pipeline demonstrates superior performance with micro F1 scores ranging from 0.94 to 0.98 on the test dataset, which is a translated version of the i2b2 2014 de-identification challenge, across 17 sensitive entities. This level of accuracy is in line with that achieved with manual de-identification by domain experts, suggesting that a fully automated and scalable process is now viable.

Automated De-Identification of Arabic Medical Records

arabicnlp-1.4

Poster

https://aclanthology.org/2023.arabicnlp-1.5.bib

https://aclanthology.org/2023.arabicnlp-1.5/

@inproceedings{toyin-etal-2023-artst, title = "{A}r{TST}: {A}rabic Text and Speech Transformer", author = "Toyin, Hawau and Djanibekov, Amirbek and Kulkarni, Ajinkya and Aldarmaki, Hanan", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.5", doi = "10.18653/v1/2023.arabicnlp-1.5", pages = "41--51", abstract = "We present ArTST, a pre-trained Arabic text and speech transformer for supporting open-source speech technologies for the Arabic language. The model architecture follows the unified-modal framework, SpeechT5, that was recently released for English, and is focused on Modern Standard Arabic (MSA), with plans to extend the model for dialectal and code-switched Arabic in future editions. We pre-trained the model from scratch on MSA speech and text data, and fine-tuned it for the following tasks: Automatic Speech Recognition (ASR), Text-To-Speech synthesis (TTS), and spoken dialect identification. In our experiments comparing ArTST with SpeechT5, as well as with previously reported results in these tasks, ArTST performs on a par with or exceeding the current state-of-the-art in all three tasks. Moreover, we find that our pre-training is conducive for generalization, which is particularly evident in the low-resource TTS task. The pre-trained model as well as the fine-tuned ASR and TTS models are released for research use.", }

We present ArTST, a pre-trained Arabic text and speech transformer for supporting open-source speech technologies for the Arabic language. The model architecture follows the unified-modal framework, SpeechT5, that was recently released for English, and is focused on Modern Standard Arabic (MSA), with plans to extend the model for dialectal and code-switched Arabic in future editions. We pre-trained the model from scratch on MSA speech and text data, and fine-tuned it for the following tasks: Automatic Speech Recognition (ASR), Text-To-Speech synthesis (TTS), and spoken dialect identification. In our experiments comparing ArTST with SpeechT5, as well as with previously reported results in these tasks, ArTST performs on a par with or exceeding the current state-of-the-art in all three tasks. Moreover, we find that our pre-training is conducive for generalization, which is particularly evident in the low-resource TTS task. The pre-trained model as well as the fine-tuned ASR and TTS models are released for research use.

ArTST: Arabic Text and Speech Transformer

arabicnlp-1.5

2310.16621

https://huggingface.co/papers/2310.16621

Poster

https://aclanthology.org/2023.arabicnlp-1.6.bib

https://aclanthology.org/2023.arabicnlp-1.6/

@inproceedings{kadaoui-etal-2023-tarjamat, title = "{TARJAMAT}: Evaluation of Bard and {C}hat{GPT} on Machine Translation of Ten {A}rabic Varieties", author = "Kadaoui, Karima and Magdy, Samar and Waheed, Abdul and Khondaker, Md Tawkat Islam and El-Shangiti, Ahmed and Nagoudi, El Moatez Billah and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.6", doi = "10.18653/v1/2023.arabicnlp-1.6", pages = "52--75", abstract = "Despite the purported multilingual proficiency of instruction-finetuned large language models (LLMs) such as ChatGPT and Bard, the linguistic inclusivity of these models remains insufficiently explored. Considering this constraint, we present a thorough assessment of Bard and ChatGPT (encompassing both GPT-3.5 and GPT-4) regarding their machine translation proficiencies across ten varieties of Arabic. Our evaluation covers diverse Arabic varieties such as Classical Arabic (CA), Modern Standard Arabic (MSA), and several country-level dialectal variants. Our analysis indicates that LLMs may encounter challenges with dialects for which minimal public datasets exist, but on average are better translators of dialects than existing commercial systems. On CA and MSA, instruction-tuned LLMs, however, trail behind commercial systems such as Google Translate. Finally, we undertake a human-centric study to scrutinize the efficacy of the relatively recent model, Bard, in following human instructions during translation tasks. Our analysis reveals a circumscribed capability of Bard in aligning with human instructions in translation contexts. Collectively, our findings underscore that prevailing LLMs remain far from inclusive, with only limited ability to cater for the linguistic and cultural intricacies of diverse communities.", }

Despite the purported multilingual proficiency of instruction-finetuned large language models (LLMs) such as ChatGPT and Bard, the linguistic inclusivity of these models remains insufficiently explored. Considering this constraint, we present a thorough assessment of Bard and ChatGPT (encompassing both GPT-3.5 and GPT-4) regarding their machine translation proficiencies across ten varieties of Arabic. Our evaluation covers diverse Arabic varieties such as Classical Arabic (CA), Modern Standard Arabic (MSA), and several country-level dialectal variants. Our analysis indicates that LLMs may encounter challenges with dialects for which minimal public datasets exist, but on average are better translators of dialects than existing commercial systems. On CA and MSA, instruction-tuned LLMs, however, trail behind commercial systems such as Google Translate. Finally, we undertake a human-centric study to scrutinize the efficacy of the relatively recent model, Bard, in following human instructions during translation tasks. Our analysis reveals a circumscribed capability of Bard in aligning with human instructions in translation contexts. Collectively, our findings underscore that prevailing LLMs remain far from inclusive, with only limited ability to cater for the linguistic and cultural intricacies of diverse communities.

TARJAMAT: Evaluation of Bard and ChatGPT on Machine Translation of Ten Arabic Varieties

arabicnlp-1.6

2308.03051

Poster

https://aclanthology.org/2023.arabicnlp-1.7.bib

https://aclanthology.org/2023.arabicnlp-1.7/

@inproceedings{pavlova-2023-leveraging, title = "Leveraging Domain Adaptation and Data Augmentation to Improve Qur{'}anic {IR} in {E}nglish and {A}rabic", author = "Pavlova, Vera", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.7", doi = "10.18653/v1/2023.arabicnlp-1.7", pages = "76--88", abstract = "In this work, we approach the problem of Qur{'}anic information retrieval (IR) in Arabic and English. Using the latest state-of-the-art methods in neural IR, we research what helps to tackle this task more efficiently. Training retrieval models requires a lot of data, which is difficult to obtain for training in-domain. Therefore, we commence with training on a large amount of general domain data and then continue training on in-domain data. To handle the lack of in-domain data, we employed a data augmentation technique, which considerably improved results in MRR@10 and NDCG@5 metrics, setting the state-of-the-art in Qur{'}anic IR for both English and Arabic. The absence of an Islamic corpus and domain-specific model for IR task in English motivated us to address this lack of resources and take preliminary steps of the Islamic corpus compilation and domain-specific language model (LM) pre-training, which helped to improve the performance of the retrieval models that use the domain-specific LM as the shared backbone. We examined several language models (LMs) in Arabic to select one that efficiently deals with the Qur{'}anic IR task. Besides transferring successful experiments from English to Arabic, we conducted additional experiments with retrieval task in Arabic to amortize the scarcity of general domain datasets used to train the retrieval models. Handling Qur{'}anic IR task combining English and Arabic allowed us to enhance the comparison and share valuable insights across models and languages.", }

In this work, we approach the problem of Qur{'}anic information retrieval (IR) in Arabic and English. Using the latest state-of-the-art methods in neural IR, we research what helps to tackle this task more efficiently. Training retrieval models requires a lot of data, which is difficult to obtain for training in-domain. Therefore, we commence with training on a large amount of general domain data and then continue training on in-domain data. To handle the lack of in-domain data, we employed a data augmentation technique, which considerably improved results in MRR@10 and NDCG@5 metrics, setting the state-of-the-art in Qur{'}anic IR for both English and Arabic. The absence of an Islamic corpus and domain-specific model for IR task in English motivated us to address this lack of resources and take preliminary steps of the Islamic corpus compilation and domain-specific language model (LM) pre-training, which helped to improve the performance of the retrieval models that use the domain-specific LM as the shared backbone. We examined several language models (LMs) in Arabic to select one that efficiently deals with the Qur{'}anic IR task. Besides transferring successful experiments from English to Arabic, we conducted additional experiments with retrieval task in Arabic to amortize the scarcity of general domain datasets used to train the retrieval models. Handling Qur{'}anic IR task combining English and Arabic allowed us to enhance the comparison and share valuable insights across models and languages.

Leveraging Domain Adaptation and Data Augmentation to Improve Qur'anic IR in English and Arabic

arabicnlp-1.7

2312.02803

https://huggingface.co/papers/2312.02803

Poster

https://aclanthology.org/2023.arabicnlp-1.8.bib

https://aclanthology.org/2023.arabicnlp-1.8/

@inproceedings{alhamadani-etal-2023-lans, title = "{LANS}: Large-scale {A}rabic News Summarization Corpus", author = "Alhamadani, Abdulaziz and Zhang, Xuchao and He, Jianfeng and Khatri, Aadyant and Lu, Chang-Tien", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.8", doi = "10.18653/v1/2023.arabicnlp-1.8", pages = "89--100", abstract = "Text summarization has been intensively studied in many languages, and some languages have reached advanced stages. Yet, Arabic Text Summarization (ATS) is still in its developing stages. Existing ATS datasets are either small or lack diversity. We build, LANS, a large-scale and diverse dataset for Arabic Text Summarization task. LANS offers 8.4 million articles and their summaries extracted from newspapers websites{'} metadata between 1999 and 2019. The high-quality and diverse summaries are written by journalists from 22 major Arab newspapers and include an eclectic mix of at least more than 7 topics from each source. We conduct an intrinsic evaluation on LANS by both automatic and human evaluations. Human evaluation of 1,000 random samples reports 95.4{\%} accuracy for our collected summaries, and automatic evaluation quantifies the diversity and abstractness of the summaries.", }

Text summarization has been intensively studied in many languages, and some languages have reached advanced stages. Yet, Arabic Text Summarization (ATS) is still in its developing stages. Existing ATS datasets are either small or lack diversity. We build, LANS, a large-scale and diverse dataset for Arabic Text Summarization task. LANS offers 8.4 million articles and their summaries extracted from newspapers websites{'} metadata between 1999 and 2019. The high-quality and diverse summaries are written by journalists from 22 major Arab newspapers and include an eclectic mix of at least more than 7 topics from each source. We conduct an intrinsic evaluation on LANS by both automatic and human evaluations. Human evaluation of 1,000 random samples reports 95.4{\%} accuracy for our collected summaries, and automatic evaluation quantifies the diversity and abstractness of the summaries.

LANS: Large-scale Arabic News Summarization Corpus

arabicnlp-1.8

2210.13600

Poster

https://aclanthology.org/2023.arabicnlp-1.9.bib

https://aclanthology.org/2023.arabicnlp-1.9/

@inproceedings{kwon-etal-2023-beyond, title = "Beyond {E}nglish: Evaluating {LLM}s for {A}rabic Grammatical Error Correction", author = "Kwon, Sang and Bhatia, Gagan and Nagoudi, El Moatez Billah and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.9", doi = "10.18653/v1/2023.arabicnlp-1.9", pages = "101--119", abstract = "Large language models (LLMs) finetuned to follow human instruction have recently exhibited significant capabilities in various English NLP tasks. However, their performance in grammatical error correction (GEC), especially on languages other than English, remains significantly unexplored. In this work, we evaluate the abilities of instruction finetuned LLMs in Arabic GEC, a complex task due to Arabic{'}s rich morphology. Our findings suggest that various prompting methods, coupled with (in-context) few-shot learning, demonstrate considerable effectiveness, with GPT-4 achieving up to 65.49 F1 score under expert prompting (approximately 5 points higher than our established baseline). Despite these positive results, we find that instruction finetuned models, regardless of their size, are still outperformed by fully finetuned ones, even if they are significantly smaller in size. This disparity highlights substantial room for improvements for LLMs. Inspired by methods used in low-resource machine translation, we also develop a method exploiting synthetic data that significantly outperforms previous models on two standard Arabic benchmarks. Our best model achieves a new SOTA on Arabic GEC, with 73.29 and 73.26 F1 on the 2014 and 2015 QALB datasets, respectively, compared to peer-reviewed published baselines.", }

Large language models (LLMs) finetuned to follow human instruction have recently exhibited significant capabilities in various English NLP tasks. However, their performance in grammatical error correction (GEC), especially on languages other than English, remains significantly unexplored. In this work, we evaluate the abilities of instruction finetuned LLMs in Arabic GEC, a complex task due to Arabic{'}s rich morphology. Our findings suggest that various prompting methods, coupled with (in-context) few-shot learning, demonstrate considerable effectiveness, with GPT-4 achieving up to 65.49 F1 score under expert prompting (approximately 5 points higher than our established baseline). Despite these positive results, we find that instruction finetuned models, regardless of their size, are still outperformed by fully finetuned ones, even if they are significantly smaller in size. This disparity highlights substantial room for improvements for LLMs. Inspired by methods used in low-resource machine translation, we also develop a method exploiting synthetic data that significantly outperforms previous models on two standard Arabic benchmarks. Our best model achieves a new SOTA on Arabic GEC, with 73.29 and 73.26 F1 on the 2014 and 2015 QALB datasets, respectively, compared to peer-reviewed published baselines.

Beyond English: Evaluating LLMs for Arabic Grammatical Error Correction

arabicnlp-1.9

2312.08400

https://huggingface.co/papers/2312.08400

Poster

https://aclanthology.org/2023.arabicnlp-1.10.bib

https://aclanthology.org/2023.arabicnlp-1.10/

@inproceedings{alkanhal-etal-2023-aswat, title = "Aswat: {A}rabic Audio Dataset for Automatic Speech Recognition Using Speech-Representation Learning", author = "Alkanhal, Lamya and Alessa, Abeer and Almahmoud, Elaf and Alaqil, Rana", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.10", doi = "10.18653/v1/2023.arabicnlp-1.10", pages = "120--127", abstract = "Recent advancements in self-supervised speech-representation learning for automatic speech recognition (ASR) approaches have significantly improved the results on many benchmarks with low-cost data labeling. In this paper, we train two self-supervised frameworks for ASR, namely wav2vec, and data2vec, in which we conduct multiple experiments and analyze their results. Furthermore, we introduce Aswat dataset, which covers multiple genres and features speakers with vocal variety. Aswat contains 732 hours of clean Arabic speech that can be used in the pretraining task for learning latent speech representations, which results in achieving a lower word error rate (WER) in Arabic ASR. We report the baseline results and achieve state-of-the-art WERs of 11.7{\%} and 10.3{\%} on Common Voice (CV) and the second round of Multi-Genre Broadcast (MGB-2) respectively, as a result of including our dataset Aswat.", }

Recent advancements in self-supervised speech-representation learning for automatic speech recognition (ASR) approaches have significantly improved the results on many benchmarks with low-cost data labeling. In this paper, we train two self-supervised frameworks for ASR, namely wav2vec, and data2vec, in which we conduct multiple experiments and analyze their results. Furthermore, we introduce Aswat dataset, which covers multiple genres and features speakers with vocal variety. Aswat contains 732 hours of clean Arabic speech that can be used in the pretraining task for learning latent speech representations, which results in achieving a lower word error rate (WER) in Arabic ASR. We report the baseline results and achieve state-of-the-art WERs of 11.7{\%} and 10.3{\%} on Common Voice (CV) and the second round of Multi-Genre Broadcast (MGB-2) respectively, as a result of including our dataset Aswat.

Aswat: Arabic Audio Dataset for Automatic Speech Recognition Using Speech-Representation Learning

arabicnlp-1.10

Poster

https://aclanthology.org/2023.arabicnlp-1.11.bib

https://aclanthology.org/2023.arabicnlp-1.11/

@inproceedings{boughorbel-hawasly-2023-analyzing, title = "Analyzing Multilingual Competency of {LLM}s in Multi-Turn Instruction Following: A Case Study of {A}rabic", author = "Boughorbel, Sabri and Hawasly, Majd", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.11", doi = "10.18653/v1/2023.arabicnlp-1.11", pages = "128--139", abstract = "While significant progress has been made in benchmarking Large Language Models (LLMs) across various tasks, there is a lack of comprehensive evaluation of their abilities in responding to multi-turn instructions in less-commonly tested languages like Arabic. Our paper offers a detailed examination of the proficiency of open LLMs in such scenarios in Arabic. Utilizing a customized Arabic translation of the MT-Bench benchmark suite, we employ GPT-4 as a uniform evaluator for both English and Arabic queries to assess and compare the performance of the LLMs on various open-ended tasks. Our findings reveal variations in model responses on different task categories, e.g., logic vs. literacy, when instructed in English or Arabic. We find that fine-tuned base models using multilingual and multi-turn datasets could be competitive to models trained from scratch on multilingual data. Finally, we hypothesize that an ensemble of small, open LLMs could perform competitively to proprietary LLMs on the benchmark.", }

While significant progress has been made in benchmarking Large Language Models (LLMs) across various tasks, there is a lack of comprehensive evaluation of their abilities in responding to multi-turn instructions in less-commonly tested languages like Arabic. Our paper offers a detailed examination of the proficiency of open LLMs in such scenarios in Arabic. Utilizing a customized Arabic translation of the MT-Bench benchmark suite, we employ GPT-4 as a uniform evaluator for both English and Arabic queries to assess and compare the performance of the LLMs on various open-ended tasks. Our findings reveal variations in model responses on different task categories, e.g., logic vs. literacy, when instructed in English or Arabic. We find that fine-tuned base models using multilingual and multi-turn datasets could be competitive to models trained from scratch on multilingual data. Finally, we hypothesize that an ensemble of small, open LLMs could perform competitively to proprietary LLMs on the benchmark.

Analyzing Multilingual Competency of LLMs in Multi-Turn Instruction Following: A Case Study of Arabic

arabicnlp-1.11

2310.14819

https://huggingface.co/papers/2310.14819

Poster

https://aclanthology.org/2023.arabicnlp-1.12.bib

https://aclanthology.org/2023.arabicnlp-1.12/

@inproceedings{elkhbir-etal-2023-cross, title = "Cross-Dialectal Named Entity Recognition in {A}rabic", author = "El Elkhbir, Niama and Zaratiana, Urchade and Tomeh, Nadi and Charnois, Thierry", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.12", doi = "10.18653/v1/2023.arabicnlp-1.12", pages = "140--149", abstract = "In this paper, we study the transferability of Named Entity Recognition (NER) models between Arabic dialects. This question is important because the available manually-annotated resources are not distributed equally across dialects: Modern Standard Arabic (MSA) is much richer than other dialects for which little to no datasets exist. How well does a NER model, trained on MSA, perform on other dialects? To answer this question, we construct four datasets. The first is an MSA dataset extracted from the ACE 2005 corpus. The others are datasets for Egyptian, Morocan and Syrian which we manually annotate following the ACE guidelines. We train a span-based NER model on top of a pretrained language model (PLM) encoder on the MSA data and study its performance on the other datasets in zero-shot settings. We study the performance of multiple PLM encoders from the literature and show that they achieve acceptable performance with no annotation effort. Our annotations and models are publicly available (\url{https://github.com/niamaelkhbir/Arabic-Cross-Dialectal-NER}).", }

In this paper, we study the transferability of Named Entity Recognition (NER) models between Arabic dialects. This question is important because the available manually-annotated resources are not distributed equally across dialects: Modern Standard Arabic (MSA) is much richer than other dialects for which little to no datasets exist. How well does a NER model, trained on MSA, perform on other dialects? To answer this question, we construct four datasets. The first is an MSA dataset extracted from the ACE 2005 corpus. The others are datasets for Egyptian, Morocan and Syrian which we manually annotate following the ACE guidelines. We train a span-based NER model on top of a pretrained language model (PLM) encoder on the MSA data and study its performance on the other datasets in zero-shot settings. We study the performance of multiple PLM encoders from the literature and show that they achieve acceptable performance with no annotation effort. Our annotations and models are publicly available (\url{https://github.com/niamaelkhbir/Arabic-Cross-Dialectal-NER}).

Cross-Dialectal Named Entity Recognition in Arabic

arabicnlp-1.12

Poster

https://aclanthology.org/2023.arabicnlp-1.13.bib

https://aclanthology.org/2023.arabicnlp-1.13/

@inproceedings{zhang-etal-2023-enhancing-arabic, title = "Enhancing {A}rabic Machine Translation for {E}-commerce Product Information: Data Quality Challenges and Innovative Selection Approaches", author = "Zhang, Bryan and Danial, Salah and Walter, Stephan", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.13", doi = "10.18653/v1/2023.arabicnlp-1.13", pages = "150--157", abstract = "Product information in e-commerce is usually localized using machine translation (MT) systems. Arabic language has rich morphology and dialectal variations, so Arabic MT in e-commerce training requires a larger volume of data from diverse data sources; Given the dynamic nature of e-commerce, such data needs to be acquired periodically to update the MT. Consequently, validating the quality of training data periodically within an industrial setting presents a notable challenge. Meanwhile, the performance of MT systems is significantly impacted by the quality and appropriateness of the training data. Hence, this study first examines the Arabic MT in e-commerce and investigates the data quality challenges for English-Arabic MT in e-commerce then proposes heuristics-based and topic-based data selection approaches to improve MT for product information. Both online and offline experiment results have shown our proposed approaches are effective, leading to improved shopping experiences for customers.", }

Product information in e-commerce is usually localized using machine translation (MT) systems. Arabic language has rich morphology and dialectal variations, so Arabic MT in e-commerce training requires a larger volume of data from diverse data sources; Given the dynamic nature of e-commerce, such data needs to be acquired periodically to update the MT. Consequently, validating the quality of training data periodically within an industrial setting presents a notable challenge. Meanwhile, the performance of MT systems is significantly impacted by the quality and appropriateness of the training data. Hence, this study first examines the Arabic MT in e-commerce and investigates the data quality challenges for English-Arabic MT in e-commerce then proposes heuristics-based and topic-based data selection approaches to improve MT for product information. Both online and offline experiment results have shown our proposed approaches are effective, leading to improved shopping experiences for customers.

Enhancing Arabic Machine Translation for E-commerce Product Information: Data Quality Challenges and Innovative Selection Approaches

arabicnlp-1.13

Poster

https://aclanthology.org/2023.arabicnlp-1.14.bib

https://aclanthology.org/2023.arabicnlp-1.14/

@inproceedings{suwaileh-etal-2023-idrisi-arabic, title = "{IDRISI}-{D}: {A}rabic and {E}nglish Datasets and Benchmarks for Location Mention Disambiguation over Disaster Microblogs", author = "Suwaileh, Reem and Elsayed, Tamer and Imran, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.14", doi = "10.18653/v1/2023.arabicnlp-1.14", pages = "158--169", abstract = "Extracting and disambiguating geolocation information from social media data enables effective disaster management, as it helps response authorities; for example, locating incidents for planning rescue activities and affected people for evacuation. Nevertheless, the dearth of resources and tools hinders the development and evaluation of Location Mention Disambiguation (LMD) models in the disaster management domain. Consequently, the LMD task is greatly understudied, especially for the low resource languages such as Arabic. To fill this gap, we introduce IDRISI-D, the largest to date English and the first Arabic public LMD datasets. Additionally, we introduce a modified hierarchical evaluation framework that offers a lenient and nuanced evaluation of LMD systems. We further benchmark IDRISI-D datasets using representative baselines and show the competitiveness of BERT-based models.", }

Extracting and disambiguating geolocation information from social media data enables effective disaster management, as it helps response authorities; for example, locating incidents for planning rescue activities and affected people for evacuation. Nevertheless, the dearth of resources and tools hinders the development and evaluation of Location Mention Disambiguation (LMD) models in the disaster management domain. Consequently, the LMD task is greatly understudied, especially for the low resource languages such as Arabic. To fill this gap, we introduce IDRISI-D, the largest to date English and the first Arabic public LMD datasets. Additionally, we introduce a modified hierarchical evaluation framework that offers a lenient and nuanced evaluation of LMD systems. We further benchmark IDRISI-D datasets using representative baselines and show the competitiveness of BERT-based models.

IDRISI-D: Arabic and English Datasets and Benchmarks for Location Mention Disambiguation over Disaster Microblogs

arabicnlp-1.14

Poster

https://aclanthology.org/2023.arabicnlp-1.15.bib

https://aclanthology.org/2023.arabicnlp-1.15/

@inproceedings{elshabrawy-etal-2023-camelparser2, title = "{C}amel{P}arser2.0: A State-of-the-Art Dependency Parser for {A}rabic", author = "Elshabrawy, Ahmed and AbuOdeh, Muhammed and Inoue, Go and Habash, Nizar", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.15", doi = "10.18653/v1/2023.arabicnlp-1.15", pages = "170--180", abstract = "We present CamelParser2.0, an open-source Python-based Arabic dependency parser targeting two popular Arabic dependency formalisms, the Columbia Arabic Treebank (CATiB), and Universal Dependencies (UD). The CamelParser2.0 pipeline handles the processing of raw text and produces tokenization, part-of-speech and rich morphological features. As part of developing CamelParser2.0, we explore many system design hyper-parameters, such as parsing model architecture and pretrained language model selection, achieving new state-of-the-art performance across diverse Arabic genres under gold and predicted tokenization settings.", }

We present CamelParser2.0, an open-source Python-based Arabic dependency parser targeting two popular Arabic dependency formalisms, the Columbia Arabic Treebank (CATiB), and Universal Dependencies (UD). The CamelParser2.0 pipeline handles the processing of raw text and produces tokenization, part-of-speech and rich morphological features. As part of developing CamelParser2.0, we explore many system design hyper-parameters, such as parsing model architecture and pretrained language model selection, achieving new state-of-the-art performance across diverse Arabic genres under gold and predicted tokenization settings.

CamelParser2.0: A State-of-the-Art Dependency Parser for Arabic

arabicnlp-1.15

Poster

https://aclanthology.org/2023.arabicnlp-1.16.bib

https://aclanthology.org/2023.arabicnlp-1.16/

@inproceedings{ali-etal-2023-gari, title = "{GARI}: Graph Attention for Relative Isomorphism of {A}rabic Word Embeddings", author = "Ali, Muhammad and Alshmrani, Maha and Qin, Jianbin and Hu, Yan and Wang, Di", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.16", doi = "10.18653/v1/2023.arabicnlp-1.16", pages = "181--190", abstract = "Bilingual Lexical Induction (BLI) is a core challenge in NLP, it relies on the relative isomorphism of individual embedding spaces. Existing attempts aimed at controlling the relative isomorphism of different embedding spaces fail to incorporate the impact of semantically related words in the model training objective. To address this, we propose GARI that combines the distributional training objectives with multiple isomorphism losses guided by the graph attention network. GARI considers the impact of semantical variations of words in order to define the relative isomorphism of the embedding spaces. Experimental evaluation using the Arabic language data set shows that GARI outperforms the existing research by improving the average P@1 by a relative score of up to 40.95{\%} and 76.80{\%} for in-domain and domain mismatch settings respectively.", }

Bilingual Lexical Induction (BLI) is a core challenge in NLP, it relies on the relative isomorphism of individual embedding spaces. Existing attempts aimed at controlling the relative isomorphism of different embedding spaces fail to incorporate the impact of semantically related words in the model training objective. To address this, we propose GARI that combines the distributional training objectives with multiple isomorphism losses guided by the graph attention network. GARI considers the impact of semantical variations of words in order to define the relative isomorphism of the embedding spaces. Experimental evaluation using the Arabic language data set shows that GARI outperforms the existing research by improving the average P@1 by a relative score of up to 40.95{\%} and 76.80{\%} for in-domain and domain mismatch settings respectively.

GARI: Graph Attention for Relative Isomorphism of Arabic Word Embeddings

arabicnlp-1.16

2310.13068

Poster

https://aclanthology.org/2023.arabicnlp-1.17.bib

https://aclanthology.org/2023.arabicnlp-1.17/

@inproceedings{alrowili-vijay-shanker-2023-artrivia, title = "{A}r{T}rivia: Harvesting {A}rabic {W}ikipedia to Build A New {A}rabic Question Answering Dataset", author = "Alrowili, Sultan and Vijay-Shanker, K", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.17", doi = "10.18653/v1/2023.arabicnlp-1.17", pages = "191--207", abstract = "We present ArTrivia, a new Arabic question-answering dataset consisting of more than 10,000 question-answer pairs along with relevant passages, covering a wide range of 18 diverse topics in Arabic. We created our dataset using a newly proposed pipeline that leverages diverse structured data sources from Arabic Wikipedia. Moreover, we conducted a comprehensive statistical analysis of ArTrivia and assessed the performance of each component in our pipeline. Additionally, we compared the performance of ArTrivia against the existing TyDi QA dataset using various experimental setups. Our analysis highlights the significance of often overlooked aspects in dataset creation, such as answer normalization, in enhancing the quality of QA datasets. Our evaluation also shows that ArTrivia presents more challenging and out-of-distribution questions to TyDi, raising questions about the feasibility of using ArTrivia as a complementary dataset to TyDi.", }

We present ArTrivia, a new Arabic question-answering dataset consisting of more than 10,000 question-answer pairs along with relevant passages, covering a wide range of 18 diverse topics in Arabic. We created our dataset using a newly proposed pipeline that leverages diverse structured data sources from Arabic Wikipedia. Moreover, we conducted a comprehensive statistical analysis of ArTrivia and assessed the performance of each component in our pipeline. Additionally, we compared the performance of ArTrivia against the existing TyDi QA dataset using various experimental setups. Our analysis highlights the significance of often overlooked aspects in dataset creation, such as answer normalization, in enhancing the quality of QA datasets. Our evaluation also shows that ArTrivia presents more challenging and out-of-distribution questions to TyDi, raising questions about the feasibility of using ArTrivia as a complementary dataset to TyDi.

ArTrivia: Harvesting Arabic Wikipedia to Build A New Arabic Question Answering Dataset

arabicnlp-1.17

Poster

https://aclanthology.org/2023.arabicnlp-1.18.bib

https://aclanthology.org/2023.arabicnlp-1.18/

@inproceedings{hakami-etal-2023-arsarcasmoji, title = "{A}r{S}arcas{M}oji Dataset: The Emoji Sentiment Roles in {A}rabic Ironic Contexts", author = "Hakami, Shatha Ali A. and Hendley, Robert and Smith, Phillip", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.18", doi = "10.18653/v1/2023.arabicnlp-1.18", pages = "208--217", abstract = "In digital communication, emoji are essential in decoding nuances such as irony, sarcasm, and humour. However, their incorporation in Arabic natural language processing (NLP) has been cautious because of the perceived complexities of the Arabic language. This paper introduces ArSarcasMoji, a dataset of 24,630 emoji-augmented texts, with 17. 5{\%} that shows irony. Through our analysis, we highlight specific emoji patterns paired with sentiment roles that denote irony in Arabic texts. The research counters prevailing notions, emphasising the importance of emoji{'}s role in understanding Arabic textual irony, and addresses their potential for accurate irony detection in Arabic digital content.", }

In digital communication, emoji are essential in decoding nuances such as irony, sarcasm, and humour. However, their incorporation in Arabic natural language processing (NLP) has been cautious because of the perceived complexities of the Arabic language. This paper introduces ArSarcasMoji, a dataset of 24,630 emoji-augmented texts, with 17. 5{\%} that shows irony. Through our analysis, we highlight specific emoji patterns paired with sentiment roles that denote irony in Arabic texts. The research counters prevailing notions, emphasising the importance of emoji{'}s role in understanding Arabic textual irony, and addresses their potential for accurate irony detection in Arabic digital content.

ArSarcasMoji Dataset: The Emoji Sentiment Roles in Arabic Ironic Contexts

arabicnlp-1.18

Poster

https://aclanthology.org/2023.arabicnlp-1.19.bib

https://aclanthology.org/2023.arabicnlp-1.19/

@inproceedings{alshahrani-etal-2023-performance, title = "Performance Implications of Using Unrepresentative Corpora in {A}rabic Natural Language Processing", author = "Alshahrani, Saied and Alshahrani, Norah and Dey, Soumyabrata and Matthews, Jeanna", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.19", doi = "10.18653/v1/2023.arabicnlp-1.19", pages = "218--231", abstract = "Wikipedia articles are a widely used source of training data for Natural Language Processing (NLP) research, particularly as corpora for low-resource languages like Arabic. However, it is essential to understand the extent to which these corpora reflect the representative contributions of native speakers, especially when many entries in a given language are directly translated from other languages or automatically generated through automated mechanisms. In this paper, we study the performance implications of using inorganic corpora that are not representative of native speakers and are generated through automated techniques such as bot generation or automated template-based translation. The case of the Arabic Wikipedia editions gives a unique case study of this since the Moroccan Arabic Wikipedia edition (ARY) is small but representative, the Egyptian Arabic Wikipedia edition (ARZ) is large but unrepresentative, and the Modern Standard Arabic Wikipedia edition (AR) is both large and more representative. We intrinsically evaluate the performance of two main NLP upstream tasks, namely word representation and language modeling, using word analogy evaluations and fill-mask evaluations using our two newly created datasets: Arab States Analogy Dataset (ASAD) and Masked Arab States Dataset (MASD). We demonstrate that for good NLP performance, we need both large and organic corpora; neither alone is sufficient. We show that producing large corpora through automated means can be a counter-productive, producing models that both perform worse and lack cultural richness and meaningful representation of the Arabic language and its native speakers.", }

Wikipedia articles are a widely used source of training data for Natural Language Processing (NLP) research, particularly as corpora for low-resource languages like Arabic. However, it is essential to understand the extent to which these corpora reflect the representative contributions of native speakers, especially when many entries in a given language are directly translated from other languages or automatically generated through automated mechanisms. In this paper, we study the performance implications of using inorganic corpora that are not representative of native speakers and are generated through automated techniques such as bot generation or automated template-based translation. The case of the Arabic Wikipedia editions gives a unique case study of this since the Moroccan Arabic Wikipedia edition (ARY) is small but representative, the Egyptian Arabic Wikipedia edition (ARZ) is large but unrepresentative, and the Modern Standard Arabic Wikipedia edition (AR) is both large and more representative. We intrinsically evaluate the performance of two main NLP upstream tasks, namely word representation and language modeling, using word analogy evaluations and fill-mask evaluations using our two newly created datasets: Arab States Analogy Dataset (ASAD) and Masked Arab States Dataset (MASD). We demonstrate that for good NLP performance, we need both large and organic corpora; neither alone is sufficient. We show that producing large corpora through automated means can be a counter-productive, producing models that both perform worse and lack cultural richness and meaningful representation of the Arabic language and its native speakers.

Performance Implications of Using Unrepresentative Corpora in Arabic Natural Language Processing

arabicnlp-1.19

Poster

https://aclanthology.org/2023.arabicnlp-1.20.bib

https://aclanthology.org/2023.arabicnlp-1.20/

@inproceedings{elmadany-etal-2023-octopus, title = "Octopus: A Multitask Model and Toolkit for {A}rabic Natural Language Generation", author = "Elmadany, AbdelRahim and Nagoudi, El Moatez Billah and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.20", doi = "10.18653/v1/2023.arabicnlp-1.20", pages = "232--243", abstract = "Understanding Arabic text and generating human-like responses is a challenging task. While many researchers have proposed models and solutions for individual problems, there is an acute shortage of a comprehensive Arabic natural language generation toolkit that is capable of handling a wide range of tasks. In this work, we present a robust Arabic text-to-text Transformer model, namely AraT5v2, methodically trained on extensive and diverse data, utilizing an extended sequence length of 2,048 tokens. We explore various pretraining strategies including unsupervised, supervised, and joint pertaining, under both single and multitask settings. Our models outperform competitive baselines with large margins. We take our work one step further by developing and publicly releasing OCTOPUS, a Python-based package and command-line toolkit tailored for eight Arabic generation tasks all exploiting a single model. We provide a link to the models and the toolkit through our public repository.", }

Understanding Arabic text and generating human-like responses is a challenging task. While many researchers have proposed models and solutions for individual problems, there is an acute shortage of a comprehensive Arabic natural language generation toolkit that is capable of handling a wide range of tasks. In this work, we present a robust Arabic text-to-text Transformer model, namely AraT5v2, methodically trained on extensive and diverse data, utilizing an extended sequence length of 2,048 tokens. We explore various pretraining strategies including unsupervised, supervised, and joint pertaining, under both single and multitask settings. Our models outperform competitive baselines with large margins. We take our work one step further by developing and publicly releasing OCTOPUS, a Python-based package and command-line toolkit tailored for eight Arabic generation tasks all exploiting a single model. We provide a link to the models and the toolkit through our public repository.

Octopus: A Multitask Model and Toolkit for Arabic Natural Language Generation

arabicnlp-1.20

2310.16127

https://huggingface.co/papers/2310.16127

Poster

https://aclanthology.org/2023.arabicnlp-1.21.bib

https://aclanthology.org/2023.arabicnlp-1.21/

@inproceedings{almazrouei-etal-2023-alghafa, title = "{A}l{G}hafa Evaluation Benchmark for {A}rabic Language Models", author = "Almazrouei, Ebtesam and Cojocaru, Ruxandra and Baldo, Michele and Malartic, Quentin and Alobeidli, Hamza and Mazzotta, Daniele and Penedo, Guilherme and Campesan, Giulia and Farooq, Mugariya and Alhammadi, Maitha and Launay, Julien and Noune, Badreddine", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.21", doi = "10.18653/v1/2023.arabicnlp-1.21", pages = "244--275", abstract = "Recent advances in the space of Arabic large language models have opened up a wealth of potential practical applications. From optimal training strategies, large scale data acquisition and continuously increasing NLP resources, the Arabic LLM landscape has improved in a very short span of time, despite being plagued by training data scarcity and limited evaluation resources compared to English. In line with contributing towards this ever-growing field, we introduce AlGhafa, a new multiple-choice evaluation benchmark for Arabic LLMs. For showcasing purposes, we train a new suite of models, including a 14 billion parameter model, the largest monolingual Arabic decoder-only model to date. We use a collection of publicly available datasets, as well as a newly introduced HandMade dataset consisting of 8 billion tokens. Finally, we explore the quantitative and qualitative toxicity of several Arabic models, comparing our models to existing public Arabic LLMs.", }

Recent advances in the space of Arabic large language models have opened up a wealth of potential practical applications. From optimal training strategies, large scale data acquisition and continuously increasing NLP resources, the Arabic LLM landscape has improved in a very short span of time, despite being plagued by training data scarcity and limited evaluation resources compared to English. In line with contributing towards this ever-growing field, we introduce AlGhafa, a new multiple-choice evaluation benchmark for Arabic LLMs. For showcasing purposes, we train a new suite of models, including a 14 billion parameter model, the largest monolingual Arabic decoder-only model to date. We use a collection of publicly available datasets, as well as a newly introduced HandMade dataset consisting of 8 billion tokens. Finally, we explore the quantitative and qualitative toxicity of several Arabic models, comparing our models to existing public Arabic LLMs.

AlGhafa Evaluation Benchmark for Arabic Language Models

arabicnlp-1.21

Poster

https://aclanthology.org/2023.arabicnlp-1.22.bib

https://aclanthology.org/2023.arabicnlp-1.22/

@inproceedings{jarrar-etal-2023-arbanking77, title = "{A}r{B}anking77: Intent Detection Neural Model and a New Dataset in Modern and Dialectical {A}rabic", author = "Jarrar, Mustafa and Birim, Ahmet and Khalilia, Mohammed and Erden, Mustafa and Ghanem, Sana", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.22", doi = "10.18653/v1/2023.arabicnlp-1.22", pages = "276--287", abstract = "This paper presents the ArBanking77, a large Arabic dataset for intent detection in the banking domain. Our dataset was arabized and localized from the original English Banking77 dataset, which consists of 13,083 queries to ArBanking77 dataset with 31,404 queries in both Modern Standard Arabic (MSA) and Palestinian dialect, with each query classified into one of the 77 classes (intents). Furthermore, we present a neural model, based on AraBERT, fine-tuned on ArBanking77, which achieved an F1-score of 0.9209 and 0.8995 on MSA and Palestinian dialect, respectively. We performed extensive experimentation in which we simulated low-resource settings, where the model is trained on a subset of the data and augmented with noisy queries to simulate colloquial terms, mistakes and misspellings found in real NLP systems, especially live chat queries. The data and the models are publicly available at https://sina.birzeit.edu/arbanking77.", }

This paper presents the ArBanking77, a large Arabic dataset for intent detection in the banking domain. Our dataset was arabized and localized from the original English Banking77 dataset, which consists of 13,083 queries to ArBanking77 dataset with 31,404 queries in both Modern Standard Arabic (MSA) and Palestinian dialect, with each query classified into one of the 77 classes (intents). Furthermore, we present a neural model, based on AraBERT, fine-tuned on ArBanking77, which achieved an F1-score of 0.9209 and 0.8995 on MSA and Palestinian dialect, respectively. We performed extensive experimentation in which we simulated low-resource settings, where the model is trained on a subset of the data and augmented with noisy queries to simulate colloquial terms, mistakes and misspellings found in real NLP systems, especially live chat queries. The data and the models are publicly available at https://sina.birzeit.edu/arbanking77.

ArBanking77: Intent Detection Neural Model and a New Dataset in Modern and Dialectical Arabic

arabicnlp-1.22

2310.19034

https://huggingface.co/papers/2310.19034

Poster

https://aclanthology.org/2023.arabicnlp-1.23.bib

https://aclanthology.org/2023.arabicnlp-1.23/

@inproceedings{zeinalipour-etal-2023-arabicros, title = "{A}rab{I}cros: {AI}-Powered {A}rabic Crossword Puzzle Generation for Educational Applications", author = "Zeinalipour, Kamyar and Saad, Mohamed and Maggini, Marco and Gori, Marco", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.23", doi = "10.18653/v1/2023.arabicnlp-1.23", pages = "288--301", abstract = "This paper presents the first Arabic crossword puzzle generator driven by advanced AI technology. Leveraging cutting-edge large language models including GPT4, GPT3-Davinci, GPT3-Curie, GPT3-Babbage, GPT3-Ada, and BERT, the system generates distinctive and challenging clues. Based on a dataset comprising over 50,000 clue-answer pairs, the generator employs fine-tuning, few/zero-shot learning strategies, and rigorous quality-checking protocols to enforce the generation of high-quality clue-answer pairs. Importantly, educational crosswords contribute to enhancing memory, expanding vocabulary, and promoting problem-solving skills, thereby augmenting the learning experience through a fun and engaging approach, reshaping the landscape of traditional learning methods. The overall system can be exploited as a powerful educational tool that amalgamates AI and innovative learning techniques, heralding a transformative era for Arabic crossword puzzles and the intersection of technology and education.", }

This paper presents the first Arabic crossword puzzle generator driven by advanced AI technology. Leveraging cutting-edge large language models including GPT4, GPT3-Davinci, GPT3-Curie, GPT3-Babbage, GPT3-Ada, and BERT, the system generates distinctive and challenging clues. Based on a dataset comprising over 50,000 clue-answer pairs, the generator employs fine-tuning, few/zero-shot learning strategies, and rigorous quality-checking protocols to enforce the generation of high-quality clue-answer pairs. Importantly, educational crosswords contribute to enhancing memory, expanding vocabulary, and promoting problem-solving skills, thereby augmenting the learning experience through a fun and engaging approach, reshaping the landscape of traditional learning methods. The overall system can be exploited as a powerful educational tool that amalgamates AI and innovative learning techniques, heralding a transformative era for Arabic crossword puzzles and the intersection of technology and education.

ArabIcros: AI-Powered Arabic Crossword Puzzle Generation for Educational Applications

arabicnlp-1.23

2312.01339

Poster

https://aclanthology.org/2023.arabicnlp-1.24.bib

https://aclanthology.org/2023.arabicnlp-1.24/

@inproceedings{al-kharusi-aalabdulsalam-2023-machine, title = "Machine Translation of {O}mani {A}rabic Dialect from Social Media", author = "Al-Kharusi, Khoula and AAlAbdulsalam, Abdurahman", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.24", doi = "10.18653/v1/2023.arabicnlp-1.24", pages = "302--309", abstract = "Research studies on Machine Translation (MT) between Modern Standard Arabic (MSA) and English are abundant. However, studies on MT between Omani Arabic (OA) dialects and English are very scarce. This research study focuses on the lack of availability of an Omani dialect parallel dataset, as well as MT of OA to English. The study uses social media data from X (formerly Twitter) to build an authentic parallel text of the Omani dialects. The research presents baseline results on this dataset using Google Translate, Microsoft Translation, and Marian NMT. A taxonomy of the most common linguistic errors is used to analyze the translations made by the NMT systems to provide insights on future improvements. Finally, transfer learning is used to adapt Marian NMT to the Omani dialect, which significantly improved by 9.88 points in the BLEU score.", }

Research studies on Machine Translation (MT) between Modern Standard Arabic (MSA) and English are abundant. However, studies on MT between Omani Arabic (OA) dialects and English are very scarce. This research study focuses on the lack of availability of an Omani dialect parallel dataset, as well as MT of OA to English. The study uses social media data from X (formerly Twitter) to build an authentic parallel text of the Omani dialects. The research presents baseline results on this dataset using Google Translate, Microsoft Translation, and Marian NMT. A taxonomy of the most common linguistic errors is used to analyze the translations made by the NMT systems to provide insights on future improvements. Finally, transfer learning is used to adapt Marian NMT to the Omani dialect, which significantly improved by 9.88 points in the BLEU score.

Machine Translation of Omani Arabic Dialect from Social Media

arabicnlp-1.24

Poster

https://aclanthology.org/2023.arabicnlp-1.25.bib

https://aclanthology.org/2023.arabicnlp-1.25/

@inproceedings{liqreina-etal-2023-arabic, title = "{A}rabic Fine-Grained Entity Recognition", author = "Liqreina, Haneen and Jarrar, Mustafa and Khalilia, Mohammed and El-Shangiti, Ahmed and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.25", doi = "10.18653/v1/2023.arabicnlp-1.25", pages = "310--323", abstract = "Traditional NER systems are typically trained to recognize coarse-grained categories of entities, and less attention is given to classifying entities into a hierarchy of fine-grained lower-level sub-types. This article aims to advance Arabic NER with fine-grained entities. We chose to extend Wojood (an open-source Nested Arabic Named Entity Corpus) with sub-types. In particular, four main entity types in Wojood (geopolitical entity (GPE), location (LOC), organization (ORG), and facility (FAC) are extended with 31 sub-types of entities. To do this, we first revised Wojood{'}s annotations of GPE, LOC, ORG, and FAC to be compatible with the LDC{'}s ACE guidelines, which yielded 5, 614 changes. Second, all mentions of GPE, LOC, ORG, and FAC ({\textasciitilde} 44K) in Wojood are manually annotated with the LDC{'}s ACE subtypes. This extended version of Wojood is called WojoodFine. To evaluate our annotations, we measured the inter-annotator agreement (IAA) using both Cohen{'}s Kappa and F1 score, resulting in 0.9861 and 0.9889, respectively. To compute the baselines of WojoodFine, we fine-tune three pre-trained Arabic BERT encoders in three settings: flat NER, nested NER and nested NER with sub-types and achieved F1 score of 0.920, 0.866, and 0.885, respectively. Our corpus and models are open source and available at https://sina.birzeit.edu/wojood/.", }

Traditional NER systems are typically trained to recognize coarse-grained categories of entities, and less attention is given to classifying entities into a hierarchy of fine-grained lower-level sub-types. This article aims to advance Arabic NER with fine-grained entities. We chose to extend Wojood (an open-source Nested Arabic Named Entity Corpus) with sub-types. In particular, four main entity types in Wojood (geopolitical entity (GPE), location (LOC), organization (ORG), and facility (FAC) are extended with 31 sub-types of entities. To do this, we first revised Wojood{'}s annotations of GPE, LOC, ORG, and FAC to be compatible with the LDC{'}s ACE guidelines, which yielded 5, 614 changes. Second, all mentions of GPE, LOC, ORG, and FAC ({\textasciitilde} 44K) in Wojood are manually annotated with the LDC{'}s ACE subtypes. This extended version of Wojood is called WojoodFine. To evaluate our annotations, we measured the inter-annotator agreement (IAA) using both Cohen{'}s Kappa and F1 score, resulting in 0.9861 and 0.9889, respectively. To compute the baselines of WojoodFine, we fine-tune three pre-trained Arabic BERT encoders in three settings: flat NER, nested NER and nested NER with sub-types and achieved F1 score of 0.920, 0.866, and 0.885, respectively. Our corpus and models are open source and available at https://sina.birzeit.edu/wojood/.

Arabic Fine-Grained Entity Recognition

arabicnlp-1.25

2310.17333

Poster

https://aclanthology.org/2023.arabicnlp-1.26.bib

https://aclanthology.org/2023.arabicnlp-1.26/

@inproceedings{alyafeai-ahmed-2023-investigating, title = "Investigating Zero-shot Cross-lingual Language Understanding for {A}rabic", author = "Alyafeai, Zaid and Ahmed, Moataz", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.26", doi = "10.18653/v1/2023.arabicnlp-1.26", pages = "324--334", abstract = "Numerous languages exhibit shared characteristics, especially in morphological features. For instance, Arabic and Russian both belong to the fusional language category. The question arises: Do such common traits influence language comprehension across diverse linguistic backgrounds? This study explores the possibility of transferring comprehension skills across languages to Arabic in a zero-shot scenario. Specifically, we demonstrate that training language models on other languages can enhance comprehension of Arabic, as evidenced by our evaluations in three key tasks: natural language inference, question answering, and named entity recognition. Our experiments reveal that certain morphologically rich languages (MRLs), such as Russian, display similarities to Arabic when assessed in a zero-shot context, particularly in tasks like question answering and natural language inference. However, this similarity is less pronounced in tasks like named entity recognition.", }

Numerous languages exhibit shared characteristics, especially in morphological features. For instance, Arabic and Russian both belong to the fusional language category. The question arises: Do such common traits influence language comprehension across diverse linguistic backgrounds? This study explores the possibility of transferring comprehension skills across languages to Arabic in a zero-shot scenario. Specifically, we demonstrate that training language models on other languages can enhance comprehension of Arabic, as evidenced by our evaluations in three key tasks: natural language inference, question answering, and named entity recognition. Our experiments reveal that certain morphologically rich languages (MRLs), such as Russian, display similarities to Arabic when assessed in a zero-shot context, particularly in tasks like question answering and natural language inference. However, this similarity is less pronounced in tasks like named entity recognition.

Investigating Zero-shot Cross-lingual Language Understanding for Arabic

arabicnlp-1.26

Poster

https://aclanthology.org/2023.arabicnlp-1.27.bib

https://aclanthology.org/2023.arabicnlp-1.27/

@inproceedings{al-thubaity-etal-2023-evaluating, title = "Evaluating {C}hat{GPT} and Bard {AI} on {A}rabic Sentiment Analysis", author = "Al-Thubaity, Abdulmohsen and Alkhereyf, Sakhar and Murayshid, Hanan and Alshalawi, Nouf and Omirah, Maha and Alateeq, Raghad and Almutairi, Rawabi and Alsuwailem, Razan and Alhassoun, Manal and Alkhanen, Imaan", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.27", doi = "10.18653/v1/2023.arabicnlp-1.27", pages = "335--349", abstract = "Large Language Models (LLMs) such as ChatGPT and Bard AI have gained much attention due to their outstanding performance on a range of NLP tasks. These models have demonstrated remarkable proficiency across various languages without the necessity for full supervision. Nevertheless, their performance in low-resource languages and dialects, like Arabic dialects in comparison to English, remains to be investigated. In this paper, we conduct a comprehensive evaluation of three LLMs for Dialectal Arabic Sentiment Analysis: namely, ChatGPT based on GPT-3.5 and GPT-4, and Bard AI. We use a Saudi dialect Twitter dataset to assess their capability in sentiment text classification and generation. For classification, we compare the performance of fully fine-tuned Arabic BERT-based models with the LLMs in few-shot settings. For data generation, we evaluate the quality of the generated new sentiment samples using human and automatic evaluation methods. The experiments reveal that GPT-4 outperforms GPT-3.5 and Bard AI in sentiment analysis classification, rivaling the top-performing fully supervised BERT-based language model. However, in terms of data generation, compared to manually annotated authentic data, these generative models often fall short in producing high-quality Dialectal Arabic text suitable for sentiment analysis.", }

Large Language Models (LLMs) such as ChatGPT and Bard AI have gained much attention due to their outstanding performance on a range of NLP tasks. These models have demonstrated remarkable proficiency across various languages without the necessity for full supervision. Nevertheless, their performance in low-resource languages and dialects, like Arabic dialects in comparison to English, remains to be investigated. In this paper, we conduct a comprehensive evaluation of three LLMs for Dialectal Arabic Sentiment Analysis: namely, ChatGPT based on GPT-3.5 and GPT-4, and Bard AI. We use a Saudi dialect Twitter dataset to assess their capability in sentiment text classification and generation. For classification, we compare the performance of fully fine-tuned Arabic BERT-based models with the LLMs in few-shot settings. For data generation, we evaluate the quality of the generated new sentiment samples using human and automatic evaluation methods. The experiments reveal that GPT-4 outperforms GPT-3.5 and Bard AI in sentiment analysis classification, rivaling the top-performing fully supervised BERT-based language model. However, in terms of data generation, compared to manually annotated authentic data, these generative models often fall short in producing high-quality Dialectal Arabic text suitable for sentiment analysis.

Evaluating ChatGPT and Bard AI on Arabic Sentiment Analysis

arabicnlp-1.27

Poster

https://aclanthology.org/2023.arabicnlp-1.28.bib

https://aclanthology.org/2023.arabicnlp-1.28/

@inproceedings{fateen-mina-2023-context, title = "In-Context Meta-Learning vs. Semantic Score-Based Similarity: A Comparative Study in {A}rabic Short Answer Grading", author = "Fateen, Menna and Mine, Tsunenori", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.28", doi = "10.18653/v1/2023.arabicnlp-1.28", pages = "350--358", abstract = "Delegating short answer grading to automated systems enhances efficiency, giving teachers more time for vital human-centered aspects of education. Studies in automatic short answer grading (ASAG) approach the problem from instance-based or reference-based perspectives. Recent studies have favored instance-based methods, but they demand substantial data for training, which is often scarce in classroom settings. This study compares both approaches using an Arabic ASAG dataset. We employ in-context meta-learning for instance-based and semantic score-based similarity for reference-based grading. Results show both methods outperform a baseline and occasionally even surpass human raters when grading unseen answers. Notably, the semantic score-based similarity approach excels in zero-shot settings, outperforming in-context meta-learning. Our work contributes insights to Arabic ASAG and introduces a prompt category classification model, leveraging GPT3.5 to augment Arabic data for improved performance.", }

Delegating short answer grading to automated systems enhances efficiency, giving teachers more time for vital human-centered aspects of education. Studies in automatic short answer grading (ASAG) approach the problem from instance-based or reference-based perspectives. Recent studies have favored instance-based methods, but they demand substantial data for training, which is often scarce in classroom settings. This study compares both approaches using an Arabic ASAG dataset. We employ in-context meta-learning for instance-based and semantic score-based similarity for reference-based grading. Results show both methods outperform a baseline and occasionally even surpass human raters when grading unseen answers. Notably, the semantic score-based similarity approach excels in zero-shot settings, outperforming in-context meta-learning. Our work contributes insights to Arabic ASAG and introduces a prompt category classification model, leveraging GPT3.5 to augment Arabic data for improved performance.

In-Context Meta-Learning vs. Semantic Score-Based Similarity: A Comparative Study in Arabic Short Answer Grading

arabicnlp-1.28

Poster

https://aclanthology.org/2023.arabicnlp-1.29.bib

https://aclanthology.org/2023.arabicnlp-1.29/

@inproceedings{jarrar-etal-2023-salma, title = "{SALMA}: {A}rabic Sense-Annotated Corpus and {WSD} Benchmarks", author = "Jarrar, Mustafa and Malaysha, Sanad and Hammouda, Tymaa and Khalilia, Mohammed", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.29", doi = "10.18653/v1/2023.arabicnlp-1.29", pages = "359--369", abstract = "SALMA, the first Arabic sense-annotated corpus, consists of {\textasciitilde}34K tokens, which are all sense-annotated. The corpus is annotated using two different sense inventories simultaneously (Modern and Ghani). SALMA novelty lies in how tokens and senses are associated. Instead of linking a token to only one intended sense, SALMA links a token to multiple senses and provides a score to each sense. A smart web-based annotation tool was developed to support scoring multiple senses against a given word. In addition to sense annotations, we also annotated the corpus using six types of named entities. The quality of our annotations was assessed using various metrics (Kappa, Linear Weighted Kappa, Quadratic Weighted Kappa, Mean Average Error, and Root Mean Square Error), which show very high inter-annotator agreement. To establish a Word Sense Disambiguation baseline using our SALMA corpus, we developed an end-to-end Word Sense Disambiguation system using Target Sense Verification. We used this system to evaluate three Target Sense Verification models available in the literature. Our best model achieved an accuracy with 84.2{\%} using Modern and 78.7{\%} using Ghani. The full corpus and the annotation tool are open-source and publicly available at https://sina.birzeit.edu/salma/.", }

SALMA, the first Arabic sense-annotated corpus, consists of {\textasciitilde}34K tokens, which are all sense-annotated. The corpus is annotated using two different sense inventories simultaneously (Modern and Ghani). SALMA novelty lies in how tokens and senses are associated. Instead of linking a token to only one intended sense, SALMA links a token to multiple senses and provides a score to each sense. A smart web-based annotation tool was developed to support scoring multiple senses against a given word. In addition to sense annotations, we also annotated the corpus using six types of named entities. The quality of our annotations was assessed using various metrics (Kappa, Linear Weighted Kappa, Quadratic Weighted Kappa, Mean Average Error, and Root Mean Square Error), which show very high inter-annotator agreement. To establish a Word Sense Disambiguation baseline using our SALMA corpus, we developed an end-to-end Word Sense Disambiguation system using Target Sense Verification. We used this system to evaluate three Target Sense Verification models available in the literature. Our best model achieved an accuracy with 84.2{\%} using Modern and 78.7{\%} using Ghani. The full corpus and the annotation tool are open-source and publicly available at https://sina.birzeit.edu/salma/.

SALMA: Arabic Sense-Annotated Corpus and WSD Benchmarks

arabicnlp-1.29

2310.19029

Poster

https://aclanthology.org/2023.arabicnlp-1.30.bib

https://aclanthology.org/2023.arabicnlp-1.30/

@inproceedings{olsen-etal-2023-arabic, title = "{A}rabic dialect identification: An in-depth error analysis on the {MADAR} parallel corpus", author = "Olsen, Helene and Touileb, Samia and Velldal, Erik", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.30", doi = "10.18653/v1/2023.arabicnlp-1.30", pages = "370--384", abstract = "This paper provides a systematic analysis and comparison of the performance of state-of-the-art models on the task of fine-grained Arabic dialect identification using the MADAR parallel corpus. We test approaches based on pre-trained transformer language models in addition to Naive Bayes models with a rich set of various features. Through a comprehensive data- and error analysis, we provide valuable insights into the strengths and weaknesses of both approaches. We discuss which dialects are more challenging to differentiate, and identify potential sources of errors. Our analysis reveals an important problem with identical sentences across dialect classes in the test set of the MADAR-26 corpus, which may confuse any classifier. We also show that none of the tested approaches captures the subtle distinctions between closely related dialects.", }

This paper provides a systematic analysis and comparison of the performance of state-of-the-art models on the task of fine-grained Arabic dialect identification using the MADAR parallel corpus. We test approaches based on pre-trained transformer language models in addition to Naive Bayes models with a rich set of various features. Through a comprehensive data- and error analysis, we provide valuable insights into the strengths and weaknesses of both approaches. We discuss which dialects are more challenging to differentiate, and identify potential sources of errors. Our analysis reveals an important problem with identical sentences across dialect classes in the test set of the MADAR-26 corpus, which may confuse any classifier. We also show that none of the tested approaches captures the subtle distinctions between closely related dialects.

Arabic dialect identification: An in-depth error analysis on the MADAR parallel corpus

arabicnlp-1.30

Poster

https://aclanthology.org/2023.arabicnlp-1.31.bib

https://aclanthology.org/2023.arabicnlp-1.31/

@inproceedings{keleg-magdy-2023-arabic, title = "{A}rabic Dialect Identification under Scrutiny: Limitations of Single-label Classification", author = "Keleg, Amr and Magdy, Walid", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.31", doi = "10.18653/v1/2023.arabicnlp-1.31", pages = "385--398", abstract = "Automatic Arabic Dialect Identification (ADI) of text has gained great popularity since it was introduced in the early 2010s. Multiple datasets were developed, and yearly shared tasks have been running since 2018. However, ADI systems are reported to fail in distinguishing between the micro-dialects of Arabic. We argue that the currently adopted framing of the ADI task as a single-label classification problem is one of the main reasons for that. We highlight the limitation of the incompleteness of the Dialect labels and demonstrate how it impacts the evaluation of ADI systems. A manual error analysis for the predictions of an ADI, performed by 7 native speakers of different Arabic dialects, revealed that $\approx$ 67{\%} of the validated errors are not true errors. Consequently, we propose framing ADI as a multi-label classification task and give recommendations for designing new ADI datasets.", }

Automatic Arabic Dialect Identification (ADI) of text has gained great popularity since it was introduced in the early 2010s. Multiple datasets were developed, and yearly shared tasks have been running since 2018. However, ADI systems are reported to fail in distinguishing between the micro-dialects of Arabic. We argue that the currently adopted framing of the ADI task as a single-label classification problem is one of the main reasons for that. We highlight the limitation of the incompleteness of the Dialect labels and demonstrate how it impacts the evaluation of ADI systems. A manual error analysis for the predictions of an ADI, performed by 7 native speakers of different Arabic dialects, revealed that $\approx$ 67{\%} of the validated errors are not true errors. Consequently, we propose framing ADI as a multi-label classification task and give recommendations for designing new ADI datasets.

Arabic Dialect Identification under Scrutiny: Limitations of Single-label Classification

arabicnlp-1.31

2310.13661

Poster

https://aclanthology.org/2023.arabicnlp-1.32.bib

https://aclanthology.org/2023.arabicnlp-1.32/

@inproceedings{albared-etal-2023-arabic, title = "{A}rabic Topic Classification in the Generative and {A}uto{ML} Era", author = "Albared, Doha and Hamoud, Hadi and Zaraket, Fadi", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.32", doi = "10.18653/v1/2023.arabicnlp-1.32", pages = "399--404", abstract = "Most recent models for Arabic topic classification leveraged fine-tuning existing pre-trained transformer models and targeted a limited number of categories. More recently, advances in automated ML and generative models introduced novel potentials for the task. While these approaches work for English, it is a question of whether they perform well for low-resourced languages; Arabic in particular. This paper presents (i) ArBoNeClass; a novel Arabic dataset with an extended 14-topic class set covering modern books from social sciences and humanities along with newspaper articles, and (ii) a set of topic classifiers built from it. We finetuned an open LLM model to build ArGTClass. We compared its performance against the best models built with Vertex AI (Google), AutoML(H2O), and AutoTrain(HuggingFace). ArGTClass outperformed the VertexAi and AutoML models and was reasonably similar to the AutoTrain model.", }

Most recent models for Arabic topic classification leveraged fine-tuning existing pre-trained transformer models and targeted a limited number of categories. More recently, advances in automated ML and generative models introduced novel potentials for the task. While these approaches work for English, it is a question of whether they perform well for low-resourced languages; Arabic in particular. This paper presents (i) ArBoNeClass; a novel Arabic dataset with an extended 14-topic class set covering modern books from social sciences and humanities along with newspaper articles, and (ii) a set of topic classifiers built from it. We finetuned an open LLM model to build ArGTClass. We compared its performance against the best models built with Vertex AI (Google), AutoML(H2O), and AutoTrain(HuggingFace). ArGTClass outperformed the VertexAi and AutoML models and was reasonably similar to the AutoTrain model.

Arabic Topic Classification in the Generative and AutoML Era

arabicnlp-1.32

Poster

https://aclanthology.org/2023.arabicnlp-1.33.bib

https://aclanthology.org/2023.arabicnlp-1.33/

@inproceedings{betka-etal-2023-enhancing, title = "On Enhancing Fine-Tuning for Pre-trained Language Models", author = "Betka, Abir and Ferhat, Zeyd and Barka, Riyadh and Boutiba, Selma and Kahhoul, Zineddine and Lakhdar, Tiar and Abdelali, Ahmed and Dahmani, Habiba", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.33", doi = "10.18653/v1/2023.arabicnlp-1.33", pages = "405--410", abstract = "The remarkable capabilities of Natural Language Models to grasp language subtleties has paved the way for their widespread adoption in diverse fields. However, adapting them for specific tasks requires the time-consuming process of fine-tuning, which consumes significant computational power and energy. Therefore, optimizing the fine-tuning time is advantageous. In this study, we propose an alternate approach that limits parameter manipulation to select layers. Our exploration led to identifying layers that offer the best trade-off between time optimization and performance preservation. We further validated this approach on multiple downstream tasks, and the results demonstrated its potential to reduce fine-tuning time by up to 50{\%} while maintaining performance within a negligible deviation of less than 5{\%}. This research showcases a promising technique for significantly improving fine-tuning efficiency without compromising task- or domain-specific learning capabilities.", }

The remarkable capabilities of Natural Language Models to grasp language subtleties has paved the way for their widespread adoption in diverse fields. However, adapting them for specific tasks requires the time-consuming process of fine-tuning, which consumes significant computational power and energy. Therefore, optimizing the fine-tuning time is advantageous. In this study, we propose an alternate approach that limits parameter manipulation to select layers. Our exploration led to identifying layers that offer the best trade-off between time optimization and performance preservation. We further validated this approach on multiple downstream tasks, and the results demonstrated its potential to reduce fine-tuning time by up to 50{\%} while maintaining performance within a negligible deviation of less than 5{\%}. This research showcases a promising technique for significantly improving fine-tuning efficiency without compromising task- or domain-specific learning capabilities.

On Enhancing Fine-Tuning for Pre-trained Language Models

arabicnlp-1.33

Poster

https://aclanthology.org/2023.arabicnlp-1.34.bib

https://aclanthology.org/2023.arabicnlp-1.34/

@inproceedings{krubinski-etal-2023-multi, title = "Multi-Parallel Corpus of {N}orth {L}evantine {A}rabic", author = "Krubi{\'n}ski, Mateusz and Sellat, Hashem and Saleh, Shadi and Posp{\'\i}{\v{s}}il, Adam and Zem{\'a}nek, Petr and Pecina, Pavel", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.34", doi = "10.18653/v1/2023.arabicnlp-1.34", pages = "411--417", abstract = "Low-resource Machine Translation (MT) is characterized by the scarce availability of training data and/or standardized evaluation benchmarks. In the context of Dialectal Arabic, recent works introduced several evaluation benchmarks covering both Modern Standard Arabic (MSA) and dialects, mapping, however, mostly to a single Indo-European language - English. In this work, we introduce a multi-lingual corpus consisting of 120,600 multi-parallel sentences in English, French, German, Greek, Spanish, and MSA selected from the OpenSubtitles corpus, which were manually translated into the North Levantine Arabic. By conducting a series of training and fine-tuning experiments, we explore how this novel resource can contribute to the research on Arabic MT.", }

Low-resource Machine Translation (MT) is characterized by the scarce availability of training data and/or standardized evaluation benchmarks. In the context of Dialectal Arabic, recent works introduced several evaluation benchmarks covering both Modern Standard Arabic (MSA) and dialects, mapping, however, mostly to a single Indo-European language - English. In this work, we introduce a multi-lingual corpus consisting of 120,600 multi-parallel sentences in English, French, German, Greek, Spanish, and MSA selected from the OpenSubtitles corpus, which were manually translated into the North Levantine Arabic. By conducting a series of training and fine-tuning experiments, we explore how this novel resource can contribute to the research on Arabic MT.

Multi-Parallel Corpus of North Levantine Arabic

arabicnlp-1.34

Poster

https://aclanthology.org/2023.arabicnlp-1.35.bib

https://aclanthology.org/2023.arabicnlp-1.35/

@inproceedings{salaheldin-sabty-2023-simplify, title = "Simplify: Automatic {A}rabic Sentence Simplification using Word Embeddings", author = "SalahEldin, Yousef and Sabty, Caroline", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.35", doi = "10.18653/v1/2023.arabicnlp-1.35", pages = "418--422", abstract = "Automatic Text Simplification (TS) involves simplifying language complexity while preserving the original meaning. The main objective of TS is to enhance the readability of complex texts, making them more accessible to a broader range of readers. This work focuses on developing a lexical text simplification system specifically for Arabic. We utilized FastText and Arabert pre-trained embedding models to create various simplification models. Our lexical approach involves a series of steps: identifying complex words, generating potential replacements, and selecting one replacement for the complex word within a sentence. We presented two main identification models: binary and multi-complexity models. We assessed the efficacy of these models by employing BERTScore to measure the similarity between the sentences generated by these models and the intended simple sentences. This comparative analysis evaluated the effectiveness of these models in accurately identifying and selecting complex words.", }

Automatic Text Simplification (TS) involves simplifying language complexity while preserving the original meaning. The main objective of TS is to enhance the readability of complex texts, making them more accessible to a broader range of readers. This work focuses on developing a lexical text simplification system specifically for Arabic. We utilized FastText and Arabert pre-trained embedding models to create various simplification models. Our lexical approach involves a series of steps: identifying complex words, generating potential replacements, and selecting one replacement for the complex word within a sentence. We presented two main identification models: binary and multi-complexity models. We assessed the efficacy of these models by employing BERTScore to measure the similarity between the sentences generated by these models and the intended simple sentences. This comparative analysis evaluated the effectiveness of these models in accurately identifying and selecting complex words.

Simplify: Automatic Arabic Sentence Simplification using Word Embeddings

arabicnlp-1.35

Poster

https://aclanthology.org/2023.arabicnlp-1.36.bib

https://aclanthology.org/2023.arabicnlp-1.36/

@inproceedings{bensalem-etal-2023-offensive, title = "Offensive Language Detection in {A}rabizi", author = "Bensalem, Imene and Mout, Meryem and Rosso, Paolo", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.36", doi = "10.18653/v1/2023.arabicnlp-1.36", pages = "423--434", abstract = "Detecting offensive language in under-resourced languages presents a significant real-world challenge for social media platforms. This paper is the first work focused on the issue of offensive language detection in Arabizi, an under-explored topic in an under-resourced form of Arabic. For the first time, a comprehensive and critical overview of the existing work on the topic is presented. In addition, we carry out experiments using different BERT-like models and show the feasibility of detecting offensive language in Arabizi with high accuracy. Throughout a thorough analysis of results, we emphasize the complexities introduced by dialect variations and out-of-domain generalization. We use in our experiments a dataset that we have constructed by leveraging existing, albeit limited, resources. To facilitate further research, we make this dataset publicly accessible to the research community.", }

Detecting offensive language in under-resourced languages presents a significant real-world challenge for social media platforms. This paper is the first work focused on the issue of offensive language detection in Arabizi, an under-explored topic in an under-resourced form of Arabic. For the first time, a comprehensive and critical overview of the existing work on the topic is presented. In addition, we carry out experiments using different BERT-like models and show the feasibility of detecting offensive language in Arabizi with high accuracy. Throughout a thorough analysis of results, we emphasize the complexities introduced by dialect variations and out-of-domain generalization. We use in our experiments a dataset that we have constructed by leveraging existing, albeit limited, resources. To facilitate further research, we make this dataset publicly accessible to the research community.

Offensive Language Detection in Arabizi

arabicnlp-1.36

Poster

https://aclanthology.org/2023.arabicnlp-1.37.bib

https://aclanthology.org/2023.arabicnlp-1.37/

@inproceedings{kulkarni-aldarmaki-2023-yet, title = "Yet Another Model for {A}rabic Dialect Identification", author = "Kulkarni, Ajinkya and Aldarmaki, Hanan", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.37", doi = "10.18653/v1/2023.arabicnlp-1.37", pages = "435--440", abstract = "In this paper, we describe a spoken Arabic dialect identification (ADI) model for Arabic that consistently outperforms previously published results on two benchmark datasets: ADI-5 and ADI-17. We explore two architectural variations: ResNet and ECAPA-TDNN, coupled with two types of acoustic features: MFCCs and features exratected from the pre-trained self-supervised model UniSpeech-SAT Large, as well as a fusion of all four variants. We find that individually, ECAPA-TDNN network outperforms ResNet, and models with UniSpeech-SAT features outperform models with MFCCs by a large margin. Furthermore, a fusion of all four variants consistently outperforms individual models. Our best models outperform previously reported results on both datasets, with accuracies of 84.7{\%} and 96.9{\%} on ADI-5 and ADI-17, respectively.", }

In this paper, we describe a spoken Arabic dialect identification (ADI) model for Arabic that consistently outperforms previously published results on two benchmark datasets: ADI-5 and ADI-17. We explore two architectural variations: ResNet and ECAPA-TDNN, coupled with two types of acoustic features: MFCCs and features exratected from the pre-trained self-supervised model UniSpeech-SAT Large, as well as a fusion of all four variants. We find that individually, ECAPA-TDNN network outperforms ResNet, and models with UniSpeech-SAT features outperform models with MFCCs by a large margin. Furthermore, a fusion of all four variants consistently outperforms individual models. Our best models outperform previously reported results on both datasets, with accuracies of 84.7{\%} and 96.9{\%} on ADI-5 and ADI-17, respectively.

Yet Another Model for Arabic Dialect Identification

arabicnlp-1.37

2310.13812

Poster

https://aclanthology.org/2023.arabicnlp-1.38.bib

https://aclanthology.org/2023.arabicnlp-1.38/

@inproceedings{waheed-etal-2023-voxarabica, title = "{V}ox{A}rabica: A Robust Dialect-Aware {A}rabic Speech Recognition System", author = "Waheed, Abdul and Talafha, Bashar and Sullivan, Peter and Elmadany, AbdelRahim and Abdul-Mageed, Muhammad", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.38", doi = "10.18653/v1/2023.arabicnlp-1.38", pages = "441--449", abstract = "Arabic is a complex language with many varieties and dialects spoken by {\textasciitilde} 450 millions all around the world. Due to the linguistic diversity and vari-ations, it is challenging to build a robust and gen-eralized ASR system for Arabic. In this work, we address this gap by developing and demoing a system, dubbed VoxArabica, for dialect identi-fication (DID) as well as automatic speech recog-nition (ASR) of Arabic. We train a wide range of models such as HuBERT (DID), Whisper, and XLS-R (ASR) in a supervised setting for Arabic DID and ASR tasks. Our DID models are trained to identify 17 different dialects in addition to MSA. We finetune our ASR models on MSA, Egyptian, Moroccan, and mixed data. Additionally, for the re-maining dialects in ASR, we provide the option to choose various models such as Whisper and MMS in a zero-shot setting. We integrate these models into a single web interface with diverse features such as audio recording, file upload, model selec-tion, and the option to raise flags for incorrect out-puts. Overall, we believe VoxArabica will be use-ful for a wide range of audiences concerned with Arabic research. Our system is currently running at https://cdce-206-12-100-168.ngrok.io/.", }

Arabic is a complex language with many varieties and dialects spoken by {\textasciitilde} 450 millions all around the world. Due to the linguistic diversity and vari-ations, it is challenging to build a robust and gen-eralized ASR system for Arabic. In this work, we address this gap by developing and demoing a system, dubbed VoxArabica, for dialect identi-fication (DID) as well as automatic speech recog-nition (ASR) of Arabic. We train a wide range of models such as HuBERT (DID), Whisper, and XLS-R (ASR) in a supervised setting for Arabic DID and ASR tasks. Our DID models are trained to identify 17 different dialects in addition to MSA. We finetune our ASR models on MSA, Egyptian, Moroccan, and mixed data. Additionally, for the re-maining dialects in ASR, we provide the option to choose various models such as Whisper and MMS in a zero-shot setting. We integrate these models into a single web interface with diverse features such as audio recording, file upload, model selec-tion, and the option to raise flags for incorrect out-puts. Overall, we believe VoxArabica will be use-ful for a wide range of audiences concerned with Arabic research. Our system is currently running at https://cdce-206-12-100-168.ngrok.io/.

VoxArabica: A Robust Dialect-Aware Arabic Speech Recognition System

arabicnlp-1.38

2310.11069

Poster

https://aclanthology.org/2023.arabicnlp-1.39.bib

https://aclanthology.org/2023.arabicnlp-1.39/

@inproceedings{al-matham-etal-2023-ksaa, title = "{KSAA}-{RD} Shared Task: {A}rabic Reverse Dictionary", author = "Al-Matham, Rawan and Alshammari, Waad and AlOsaimy, Abdulrahman and Alhumoud, Sarah and Wazrah, Asma and Altamimi, Afrah and Alharbi, Halah and Alaifi, Abdullah", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.39", doi = "10.18653/v1/2023.arabicnlp-1.39", pages = "450--460", abstract = "This paper outlines the KSAA-RD shared task, which aims to develop a Reverse Dictionary (RD) system for the Arabic language. RDs allow users to find words based on their meanings or definition. This shared task, KSAA-RD, includes two subtasks: Arabic RD and cross-lingual reverse dictionaries (CLRD). Given a definition (referred to as a {``}gloss{''}) in either Arabic or English, the teams compete to find the most similar word embeddings of their corresponding word. The winning team achieved 24.20 and 12.70 for RD and CLRD, respectively in terms of rank metric. In this paper, we describe the methods employed by the participating teams and offer an outlook for KSAA-RD.", }

This paper outlines the KSAA-RD shared task, which aims to develop a Reverse Dictionary (RD) system for the Arabic language. RDs allow users to find words based on their meanings or definition. This shared task, KSAA-RD, includes two subtasks: Arabic RD and cross-lingual reverse dictionaries (CLRD). Given a definition (referred to as a {``}gloss{''}) in either Arabic or English, the teams compete to find the most similar word embeddings of their corresponding word. The winning team achieved 24.20 and 12.70 for RD and CLRD, respectively in terms of rank metric. In this paper, we describe the methods employed by the participating teams and offer an outlook for KSAA-RD.

KSAA-RD Shared Task: Arabic Reverse Dictionary

arabicnlp-1.39

Poster

https://aclanthology.org/2023.arabicnlp-1.40.bib

https://aclanthology.org/2023.arabicnlp-1.40/

@inproceedings{taylor-2023-uwb, title = "{UWB} at {A}rabic Reverse Dictionary shared task: Computing the meaning of a gloss", author = "Taylor, Stephen", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.40", doi = "10.18653/v1/2023.arabicnlp-1.40", pages = "461--466", abstract = "To extract the {`}meaning{'} of a gloss phrase, we build a list of sense-IDs for each word in the phrase which is in our vocabulary. We choose one sense-ID from each list so as to maximise similarity of all the IDs in the chosen subset. We take the meaning of the phrase in semantic space to be the weighted sum of the embedding vectors of the IDs.", }

To extract the {`}meaning{'} of a gloss phrase, we build a list of sense-IDs for each word in the phrase which is in our vocabulary. We choose one sense-ID from each list so as to maximise similarity of all the IDs in the chosen subset. We take the meaning of the phrase in semantic space to be the weighted sum of the embedding vectors of the IDs.

UWB at Arabic Reverse Dictionary shared task: Computing the meaning of a gloss

arabicnlp-1.40

Poster

https://aclanthology.org/2023.arabicnlp-1.41.bib

https://aclanthology.org/2023.arabicnlp-1.41/

@inproceedings{sibaee-etal-2023-qamosy, title = "Qamosy at {A}rabic Reverse Dictionary shared task: Semi Decoder Architecture for Reverse Dictionary with {SBERT} Encoder", author = "Sibaee, Serry and Ahmad, Samar and Khurfan, Ibrahim and Sabeeh, Vian and Bahaaulddin, Ahmed and Belhaj, Hanan and Alharbi, Abdullah", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.41", doi = "10.18653/v1/2023.arabicnlp-1.41", pages = "467--471", abstract = "A reverse dictionary takes a descriptive phrase of a particular concept and returns words with definitions that align with that phrase. While many reverse dictionaries cater to languages such as English and are readily available online or have been developed by researchers, there is a notable lack of similar resources for the Arabic language. This paper describes our participation in the Arabic Reverse Dictionary shared task. Our proposed method consists of two main steps: First, we convert word definitions into multidimensional vectors. Then, we train these encoded vectors using the Semi-Decoder model for our target task. Our system secured 2nd place based on the Rank metric for both embeddings (Electra and Sgns).", }

A reverse dictionary takes a descriptive phrase of a particular concept and returns words with definitions that align with that phrase. While many reverse dictionaries cater to languages such as English and are readily available online or have been developed by researchers, there is a notable lack of similar resources for the Arabic language. This paper describes our participation in the Arabic Reverse Dictionary shared task. Our proposed method consists of two main steps: First, we convert word definitions into multidimensional vectors. Then, we train these encoded vectors using the Semi-Decoder model for our target task. Our system secured 2nd place based on the Rank metric for both embeddings (Electra and Sgns).

Qamosy at Arabic Reverse Dictionary shared task: Semi Decoder Architecture for Reverse Dictionary with SBERT Encoder

arabicnlp-1.41

Poster

https://aclanthology.org/2023.arabicnlp-1.42.bib

https://aclanthology.org/2023.arabicnlp-1.42/

@inproceedings{qaddoumi-2023-abed, title = "Abed at {KSAA}-{RD} Shared Task: Enhancing {A}rabic Word Embedding with Modified {BERT} Multilingual", author = "Qaddoumi, Abdelrahim", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.42", doi = "10.18653/v1/2023.arabicnlp-1.42", pages = "472--476", abstract = "This paper presents a novel approach to the Arabic Reverse Dictionary Shared Task at WANLP 2023 by leveraging the BERT Multilingual model and introducing modifications augmentation and using a multi attention head. The proposed method aims to enhance the performance of the model in understanding and generating word embeddings for Arabic definitions, both in monolingual and cross-lingual contexts. It achieved good results compared to benchmark and other models in the shared task 1 and 2.", }

This paper presents a novel approach to the Arabic Reverse Dictionary Shared Task at WANLP 2023 by leveraging the BERT Multilingual model and introducing modifications augmentation and using a multi attention head. The proposed method aims to enhance the performance of the model in understanding and generating word embeddings for Arabic definitions, both in monolingual and cross-lingual contexts. It achieved good results compared to benchmark and other models in the shared task 1 and 2.

Abed at KSAA-RD Shared Task: Enhancing Arabic Word Embedding with Modified BERT Multilingual

arabicnlp-1.42

Poster

https://aclanthology.org/2023.arabicnlp-1.43.bib

https://aclanthology.org/2023.arabicnlp-1.43/

@inproceedings{elbakry-etal-2023-rosetta, title = "Rosetta Stone at {KSAA}-{RD} Shared Task: A Hop From Language Modeling To Word{--}Definition Alignment", author = "Elbakry, Ahmed and Gabr, Mohamed and ElNokrashy, Muhammad and AlKhamissi, Badr", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.43", doi = "10.18653/v1/2023.arabicnlp-1.43", pages = "477--482", abstract = "A Reverse Dictionary is a tool enabling users to discover a word based on its provided definition, meaning, or description. Such a technique proves valuable in various scenarios, aiding language learners who possess a description of a word without its identity, and benefiting writers seeking precise terminology. These scenarios often encapsulate what is referred to as the {``}Tip-of-the-Tongue{''} (TOT) phenomena. In this work, we present our winning solution for the Arabic Reverse Dictionary shared task. This task focuses on deriving a vector representation of an Arabic word from its accompanying description. The shared task encompasses two distinct subtasks: the first involves an Arabic definition as input, while the second employs an English definition. For the first subtask, our approach relies on an ensemble of finetuned Arabic BERT-based models, predicting the word embedding for a given definition. The final representation is obtained through averaging the output embeddings from each model within the ensemble. In contrast, the most effective solution for the second subtask involves translating the English test definitions into Arabic and applying them to the finetuned models originally trained for the first subtask. This straightforward method achieves the highest score across both subtasks.", }

A Reverse Dictionary is a tool enabling users to discover a word based on its provided definition, meaning, or description. Such a technique proves valuable in various scenarios, aiding language learners who possess a description of a word without its identity, and benefiting writers seeking precise terminology. These scenarios often encapsulate what is referred to as the {``}Tip-of-the-Tongue{''} (TOT) phenomena. In this work, we present our winning solution for the Arabic Reverse Dictionary shared task. This task focuses on deriving a vector representation of an Arabic word from its accompanying description. The shared task encompasses two distinct subtasks: the first involves an Arabic definition as input, while the second employs an English definition. For the first subtask, our approach relies on an ensemble of finetuned Arabic BERT-based models, predicting the word embedding for a given definition. The final representation is obtained through averaging the output embeddings from each model within the ensemble. In contrast, the most effective solution for the second subtask involves translating the English test definitions into Arabic and applying them to the finetuned models originally trained for the first subtask. This straightforward method achieves the highest score across both subtasks.

Rosetta Stone at KSAA-RD Shared Task: A Hop From Language Modeling To Word–Definition Alignment

arabicnlp-1.43

Poster

https://aclanthology.org/2023.arabicnlp-1.44.bib

https://aclanthology.org/2023.arabicnlp-1.44/

@inproceedings{hasanain-etal-2023-araieval, title = "{A}r{AIE}val Shared Task: Persuasion Techniques and Disinformation Detection in {A}rabic Text", author = "Hasanain, Maram and Alam, Firoj and Mubarak, Hamdy and Abdaljalil, Samir and Zaghouani, Wajdi and Nakov, Preslav and Da San Martino, Giovanni and Freihat, Abed", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.44", doi = "10.18653/v1/2023.arabicnlp-1.44", pages = "483--493", abstract = "We present an overview of the ArAIEval shared task, organized as part of the first ArabicNLP 2023 conference co-located with EMNLP 2023. ArAIEval offers two tasks over Arabic text: (1) persuasion technique detection, focusing on identifying persuasion techniques in tweets and news articles, and (2) disinformation detection in binary and multiclass setups over tweets. A total of 20 teams participated in the final evaluation phase, with 14 and 16 teams participating in Task 1 and Task 2, respectively. Across both tasks, we observe that fine-tuning transformer models such as AraBERT is the core of majority of participating systems. We provide a description of the task setup, including description of datasets construction and the evaluation setup. We also provide a brief overview of the participating systems. All datasets and evaluation scripts from the shared task are released to the research community. We hope this will enable further research on such important tasks within the Arabic NLP community.", }

We present an overview of the ArAIEval shared task, organized as part of the first ArabicNLP 2023 conference co-located with EMNLP 2023. ArAIEval offers two tasks over Arabic text: (1) persuasion technique detection, focusing on identifying persuasion techniques in tweets and news articles, and (2) disinformation detection in binary and multiclass setups over tweets. A total of 20 teams participated in the final evaluation phase, with 14 and 16 teams participating in Task 1 and Task 2, respectively. Across both tasks, we observe that fine-tuning transformer models such as AraBERT is the core of majority of participating systems. We provide a description of the task setup, including description of datasets construction and the evaluation setup. We also provide a brief overview of the participating systems. All datasets and evaluation scripts from the shared task are released to the research community. We hope this will enable further research on such important tasks within the Arabic NLP community.

ArAIEval Shared Task: Persuasion Techniques and Disinformation Detection in Arabic Text

arabicnlp-1.44

2311.03179

Poster

https://aclanthology.org/2023.arabicnlp-1.45.bib

https://aclanthology.org/2023.arabicnlp-1.45/

@inproceedings{tuck-etal-2023-detectiveredasers, title = "{D}etective{R}edasers at {A}r{AIE}val Shared Task: Leveraging Transformer Ensembles for {A}rabic Deception Detection", author = "Tuck, Bryan and Qachfar, Fatima Zahra and Boumber, Dainis and Verma, Rakesh", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.45", doi = "10.18653/v1/2023.arabicnlp-1.45", pages = "494--501", abstract = "This paper outlines a methodology aimed at combating disinformation in Arabic social media, a strategy that secured a first-place finish in tasks 2A and 2B at the ArAIEval shared task during the ArabicNLP 2023 conference. Our team, DetectiveRedasers, developed a hyperparameter-optimized pipeline centered around singular BERT-based models for the Arabic language, enhanced by a soft-voting ensemble strategy. Subsequent evaluation on the test dataset reveals that ensembles, although generally resilient, do not always outperform individual models. The primary contributions of this paper are its multifaceted strategy, which led to winning solutions for both binary (2A) and multiclass (2B) disinformation classification tasks.", }

This paper outlines a methodology aimed at combating disinformation in Arabic social media, a strategy that secured a first-place finish in tasks 2A and 2B at the ArAIEval shared task during the ArabicNLP 2023 conference. Our team, DetectiveRedasers, developed a hyperparameter-optimized pipeline centered around singular BERT-based models for the Arabic language, enhanced by a soft-voting ensemble strategy. Subsequent evaluation on the test dataset reveals that ensembles, although generally resilient, do not always outperform individual models. The primary contributions of this paper are its multifaceted strategy, which led to winning solutions for both binary (2A) and multiclass (2B) disinformation classification tasks.

DetectiveRedasers at ArAIEval Shared Task: Leveraging Transformer Ensembles for Arabic Deception Detection

arabicnlp-1.45

Poster

https://aclanthology.org/2023.arabicnlp-1.46.bib

https://aclanthology.org/2023.arabicnlp-1.46/

@inproceedings{hadjer-bouklouha-2023-hte, title = "{HTE} at {A}r{AIE}val Shared Task: Integrating Content Type Information in Binary Persuasive Technique Detection", author = "Hadjer, Khaldi and Bouklouha, Taqiy", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.46", doi = "10.18653/v1/2023.arabicnlp-1.46", pages = "502--507", abstract = "Propaganda frequently employs sophisticated persuasive strategies in order to influence public opinion and manipulate perceptions. As a result, automating the detection of persuasive techniques is critical in identifying and mitigating propaganda on social media and in mainstream media. This paper proposes a set of transformer-based models for detecting persuasive techniques in tweets and news that incorporate content type information as extra features or as an extra learning objective in a multitask learning setting. In addition to learning to detect the presence of persuasive techniques in text, our best model learns specific syntactic and lexical cues used to express them based on text genre (type) as an auxiliary task. To optimize the model and deal with data imbalance, a focal loss is used. As part of ArabicNLP2023-ArAIEval shared task, this model achieves the highest score in the shared task 1A out of 13 participants, according to the official results, with a micro-F1 of 76.34{\%} and a macro-F1 of 73.21{\%} on the test dataset.", }

Propaganda frequently employs sophisticated persuasive strategies in order to influence public opinion and manipulate perceptions. As a result, automating the detection of persuasive techniques is critical in identifying and mitigating propaganda on social media and in mainstream media. This paper proposes a set of transformer-based models for detecting persuasive techniques in tweets and news that incorporate content type information as extra features or as an extra learning objective in a multitask learning setting. In addition to learning to detect the presence of persuasive techniques in text, our best model learns specific syntactic and lexical cues used to express them based on text genre (type) as an auxiliary task. To optimize the model and deal with data imbalance, a focal loss is used. As part of ArabicNLP2023-ArAIEval shared task, this model achieves the highest score in the shared task 1A out of 13 participants, according to the official results, with a micro-F1 of 76.34{\%} and a macro-F1 of 73.21{\%} on the test dataset.

HTE at ArAIEval Shared Task: Integrating Content Type Information in Binary Persuasive Technique Detection

arabicnlp-1.46

Poster

https://aclanthology.org/2023.arabicnlp-1.47.bib

https://aclanthology.org/2023.arabicnlp-1.47/

@inproceedings{lichouri-etal-2023-usthb, title = "{USTHB} at {A}r{AIE}val{'}23 Shared Task: Disinformation Detection System based on Linguistic Feature Concatenation", author = "Lichouri, Mohamed and Lounnas, Khaled and Zitouni, Aicha and Latrache, Houda and Djeradi, Rachida", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.47", doi = "10.18653/v1/2023.arabicnlp-1.47", pages = "508--512", abstract = "In this research paper, we undertake a comprehensive examination of several pivotal factors that impact the performance of Arabic Disinformation Detection in the ArAIEval{'}2023 shared task. Our exploration encompasses the influence of surface preprocessing, morphological preprocessing, the FastText vector model, and the weighted fusion of TF-IDF features. To carry out classification tasks, we employ the Linear Support Vector Classification (LSVC) model. In the evaluation phase, our system showcases significant results, achieving an F$_1$ micro score of 76.70{\%} and 50.46{\%} for binary and multiple classification scenarios, respectively. These accomplishments closely correspond to the average F$_1$ micro scores achieved by other systems submitted for the second subtask, standing at 77.96{\%} and 64.85{\%} for binary and multiple classification scenarios, respectively.", }

In this research paper, we undertake a comprehensive examination of several pivotal factors that impact the performance of Arabic Disinformation Detection in the ArAIEval{'}2023 shared task. Our exploration encompasses the influence of surface preprocessing, morphological preprocessing, the FastText vector model, and the weighted fusion of TF-IDF features. To carry out classification tasks, we employ the Linear Support Vector Classification (LSVC) model. In the evaluation phase, our system showcases significant results, achieving an F$_1$ micro score of 76.70{\%} and 50.46{\%} for binary and multiple classification scenarios, respectively. These accomplishments closely correspond to the average F$_1$ micro scores achieved by other systems submitted for the second subtask, standing at 77.96{\%} and 64.85{\%} for binary and multiple classification scenarios, respectively.

USTHB at ArAIEval'23 Shared Task: Disinformation Detection System based on Linguistic Feature Concatenation

arabicnlp-1.47

Poster

https://aclanthology.org/2023.arabicnlp-1.48.bib

https://aclanthology.org/2023.arabicnlp-1.48/

@inproceedings{mangalvedhekar-etal-2023-mavericks, title = "Mavericks at {A}r{AIE}val Shared Task: Towards a Safer Digital Space - Transformer Ensemble Models Tackling Deception and Persuasion", author = "Mangalvedhekar, Sudeep and Deshpande, Kshitij and Patwardhan, Yash and Deshpande, Vedant and Murumkar, Ravindra", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.48", doi = "10.18653/v1/2023.arabicnlp-1.48", pages = "513--518", abstract = "In this paper, we highlight our approach for the {``}Arabic AI Tasks Evaluation (ArAiEval) Shared Task 2023{''}. We present our approaches for task 1-A and task 2-A of the shared task which focus on persuasion technique detection and disinformation detection respectively. Detection of persuasion techniques and disinformation has become imperative to avoid distortion of authentic information. The tasks use multigenre snippets of tweets and news articles for the given binary classification problem. We experiment with several transformer-based models that were pre-trained on the Arabic language. We fine-tune these state-of-the-art models on the provided dataset. Ensembling is employed to enhance the performance of the systems. We achieved a micro F1-score of 0.742 on task 1-A (8th rank on the leaderboard) and 0.901 on task 2-A (7th rank on the leaderboard) respectively.", }

In this paper, we highlight our approach for the {``}Arabic AI Tasks Evaluation (ArAiEval) Shared Task 2023{''}. We present our approaches for task 1-A and task 2-A of the shared task which focus on persuasion technique detection and disinformation detection respectively. Detection of persuasion techniques and disinformation has become imperative to avoid distortion of authentic information. The tasks use multigenre snippets of tweets and news articles for the given binary classification problem. We experiment with several transformer-based models that were pre-trained on the Arabic language. We fine-tune these state-of-the-art models on the provided dataset. Ensembling is employed to enhance the performance of the systems. We achieved a micro F1-score of 0.742 on task 1-A (8th rank on the leaderboard) and 0.901 on task 2-A (7th rank on the leaderboard) respectively.

Mavericks at ArAIEval Shared Task: Towards a Safer Digital Space - Transformer Ensemble Models Tackling Deception and Persuasion

arabicnlp-1.48

Poster

https://aclanthology.org/2023.arabicnlp-1.49.bib

https://aclanthology.org/2023.arabicnlp-1.49/

@inproceedings{veeramani-etal-2023-knowtellconvince, title = "{K}now{T}ell{C}onvince at {A}r{AIE}val Shared Task: Disinformation and Persuasion Detection in {A}rabic using Similar and Contrastive Representation Alignment", author = "Veeramani, Hariram and Thapa, Surendrabikram and Naseem, Usman", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.49", doi = "10.18653/v1/2023.arabicnlp-1.49", pages = "519--524", abstract = "In an era of widespread digital communication, the challenge of identifying and countering disinformation has become increasingly critical. However, compared to the solutions available in the English language, the resources and strategies for tackling this multifaceted problem in Arabic are relatively scarce. To address this issue, this paper presents our solutions to tasks in ArAIEval 2023. Task 1 focuses on detecting persuasion techniques, while Task 2 centers on disinformation detection within Arabic text. Leveraging a multi-head model architecture, fine-tuning techniques, sequential learning, and innovative activation functions, our contributions significantly enhance persuasion techniques and disinformation detection accuracy. Beyond improving performance, our work fills a critical research gap in content analysis for Arabic, empowering individuals, communities, and digital platforms to combat deceptive content effectively and preserve the credibility of information sources within the Arabic-speaking world.", }

In an era of widespread digital communication, the challenge of identifying and countering disinformation has become increasingly critical. However, compared to the solutions available in the English language, the resources and strategies for tackling this multifaceted problem in Arabic are relatively scarce. To address this issue, this paper presents our solutions to tasks in ArAIEval 2023. Task 1 focuses on detecting persuasion techniques, while Task 2 centers on disinformation detection within Arabic text. Leveraging a multi-head model architecture, fine-tuning techniques, sequential learning, and innovative activation functions, our contributions significantly enhance persuasion techniques and disinformation detection accuracy. Beyond improving performance, our work fills a critical research gap in content analysis for Arabic, empowering individuals, communities, and digital platforms to combat deceptive content effectively and preserve the credibility of information sources within the Arabic-speaking world.

KnowTellConvince at ArAIEval Shared Task: Disinformation and Persuasion Detection in Arabic using Similar and Contrastive Representation Alignment

arabicnlp-1.49

Poster

https://aclanthology.org/2023.arabicnlp-1.50.bib

https://aclanthology.org/2023.arabicnlp-1.50/

@inproceedings{jaber-martinez-2023-ptuk, title = "{PTUK}-{HULAT} at {A}r{AIE}val Shared Task Fine-tuned Distilbert to Predict Disinformative Tweets", author = "Jaber, Areej and Martinez, Paloma", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.50", doi = "10.18653/v1/2023.arabicnlp-1.50", pages = "525--529", abstract = "Disinformation involves the dissemination of incomplete, inaccurate, or misleading information; it has the objective, goal, or purpose of deliberately or intentionally lying to others aboutthe truth. The spread of disinformative information on social media has serious implications, and it causes concern among internet users in different aspects. Automatic classification models are required to detect disinformative posts on social media, especially on Twitter. In this article, DistilBERT multilingual model was fine-tuned to classify tweets either as dis-informative or not dis-informative in Subtask 2A of the ArAIEval shared task. The system outperformed the baseline and achieved F1 micro 87{\%} and F1 macro 80{\%}. Our system ranked 11 compared with all participants.", }

Disinformation involves the dissemination of incomplete, inaccurate, or misleading information; it has the objective, goal, or purpose of deliberately or intentionally lying to others aboutthe truth. The spread of disinformative information on social media has serious implications, and it causes concern among internet users in different aspects. Automatic classification models are required to detect disinformative posts on social media, especially on Twitter. In this article, DistilBERT multilingual model was fine-tuned to classify tweets either as dis-informative or not dis-informative in Subtask 2A of the ArAIEval shared task. The system outperformed the baseline and achieved F1 micro 87{\%} and F1 macro 80{\%}. Our system ranked 11 compared with all participants.

PTUK-HULAT at ArAIEval Shared Task Fine-tuned Distilbert to Predict Disinformative Tweets

arabicnlp-1.50

Poster

https://aclanthology.org/2023.arabicnlp-1.51.bib

https://aclanthology.org/2023.arabicnlp-1.51/

@inproceedings{bahaaulddin-etal-2023-aradetector, title = "{A}ra{D}etector at {A}r{AIE}val Shared Task: An Ensemble of {A}rabic-specific pre-trained {BERT} and {GPT}-4 for {A}rabic Disinformation Detection", author = "Bahaaulddin, Ahmed and Sabeeh, Vian and Belhaj, Hanan and Sibaee, Serry and Ahmad, Samar and Khurfan, Ibrahim and Alharbi, Abdullah", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.51", doi = "10.18653/v1/2023.arabicnlp-1.51", pages = "530--535", abstract = "The rapid proliferation of disinformation through social media has become one of the most dangerous means to deceive and influence people{'}s thoughts, viewpoints, or behaviors due to social media{'}s facilities, such as rapid access, lower cost, and ease of use. Disinformation can spread through social media in different ways, such as fake news stories, doctored images or videos, deceptive data, and even conspiracy theories, thus making detecting disinformation challenging. This paper is a part of participation in the ArAIEval competition that relates to disinformation detection. This work evaluated four models: MARBERT, the proposed ensemble model, and two tests over GPT-4 (zero-shot and Few-shot). GPT-4 achieved micro-F1 79.01{\%} while the ensemble method obtained 76.83{\%}. Despite no improvement in the micro-F1 score on the dev dataset using the ensemble approach, we still used it for the test dataset predictions. We believed that merging different classifiers might enhance the system{'}s prediction accuracy.", }

The rapid proliferation of disinformation through social media has become one of the most dangerous means to deceive and influence people{'}s thoughts, viewpoints, or behaviors due to social media{'}s facilities, such as rapid access, lower cost, and ease of use. Disinformation can spread through social media in different ways, such as fake news stories, doctored images or videos, deceptive data, and even conspiracy theories, thus making detecting disinformation challenging. This paper is a part of participation in the ArAIEval competition that relates to disinformation detection. This work evaluated four models: MARBERT, the proposed ensemble model, and two tests over GPT-4 (zero-shot and Few-shot). GPT-4 achieved micro-F1 79.01{\%} while the ensemble method obtained 76.83{\%}. Despite no improvement in the micro-F1 score on the dev dataset using the ensemble approach, we still used it for the test dataset predictions. We believed that merging different classifiers might enhance the system{'}s prediction accuracy.

AraDetector at ArAIEval Shared Task: An Ensemble of Arabic-specific pre-trained BERT and GPT-4 for Arabic Disinformation Detection

arabicnlp-1.51

Poster

https://aclanthology.org/2023.arabicnlp-1.52.bib

https://aclanthology.org/2023.arabicnlp-1.52/

@inproceedings{abdel-salam-2023-rematchka, title = "rematchka at {A}r{AIE}val Shared Task: Prefix-Tuning {\&} Prompt-tuning for Improved Detection of Propaganda and Disinformation in {A}rabic Social Media Content", author = "Abdel-Salam, Reem", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.52", doi = "10.18653/v1/2023.arabicnlp-1.52", pages = "536--542", abstract = "The rise of propaganda and disinformation in the digital age has necessitated the development of effective detection methods to combat the spread of deceptive information. In this paper we present our approach proposed for ArAIEval shared task : propaganda and disinformation detection in Arabic text. Our system utilised different pre-trained BERT based models, that makes use of prompt-learning based on knowledgeable expansion and prefix-tuning. The proposed approach secured third place in subtask-1A with 0.7555 F1-micro score, second place in subtask-1B with 0.5658 F1-micro score. However, for subtask-2A {\&} 2B, the proposed system achieved fourth place with an F1-micro score of 0.9040, 0.8219 respectively. Our findings suggest that prompt-tuning-based {\&} prefix-tuning based models performed better than conventional fine-tuning. Furthermore, using loss aware class imbalance, improved performance.", }

The rise of propaganda and disinformation in the digital age has necessitated the development of effective detection methods to combat the spread of deceptive information. In this paper we present our approach proposed for ArAIEval shared task : propaganda and disinformation detection in Arabic text. Our system utilised different pre-trained BERT based models, that makes use of prompt-learning based on knowledgeable expansion and prefix-tuning. The proposed approach secured third place in subtask-1A with 0.7555 F1-micro score, second place in subtask-1B with 0.5658 F1-micro score. However, for subtask-2A {\&} 2B, the proposed system achieved fourth place with an F1-micro score of 0.9040, 0.8219 respectively. Our findings suggest that prompt-tuning-based {\&} prefix-tuning based models performed better than conventional fine-tuning. Furthermore, using loss aware class imbalance, improved performance.

rematchka at ArAIEval Shared Task: Prefix-Tuning & Prompt-tuning for Improved Detection of Propaganda and Disinformation in Arabic Social Media Content

arabicnlp-1.52

Poster

https://aclanthology.org/2023.arabicnlp-1.53.bib

https://aclanthology.org/2023.arabicnlp-1.53/

@inproceedings{oumer-etal-2023-itri, title = "Itri Amigos at {A}r{AIE}val Shared Task: Transformer vs. Compression-Based Models for Persuasion Techniques and Disinformation Detection", author = "Oumer, Jehad and Ahmed, Nouman and Flechas Manrique, Natalia", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.53", doi = "10.18653/v1/2023.arabicnlp-1.53", pages = "543--548", abstract = "Social media has significantly amplified the dissemination of misinformation. Researchers have employed natural language processing and machine learning techniques to identify and categorize false information on these platforms. While there is a well-established body of research on detecting fake news in English and Latin languages, the study of Arabic fake news detection remains limited. This paper describes the methods used to tackle the challenges of the ArAIEval shared Task 2023. We conducted experiments with both monolingual Arabic and multi-lingual pre-trained Language Models (LM). We found that the monolingual Arabic models outperformed in all four subtasks. Additionally, we explored a novel lossless compression method, which, while not surpassing pretrained LM performance, presents an intriguing avenue for future experimentation to achieve comparable results in a more efficient and rapid manner.", }

Social media has significantly amplified the dissemination of misinformation. Researchers have employed natural language processing and machine learning techniques to identify and categorize false information on these platforms. While there is a well-established body of research on detecting fake news in English and Latin languages, the study of Arabic fake news detection remains limited. This paper describes the methods used to tackle the challenges of the ArAIEval shared Task 2023. We conducted experiments with both monolingual Arabic and multi-lingual pre-trained Language Models (LM). We found that the monolingual Arabic models outperformed in all four subtasks. Additionally, we explored a novel lossless compression method, which, while not surpassing pretrained LM performance, presents an intriguing avenue for future experimentation to achieve comparable results in a more efficient and rapid manner.

Itri Amigos at ArAIEval Shared Task: Transformer vs. Compression-Based Models for Persuasion Techniques and Disinformation Detection

arabicnlp-1.53

Poster

https://aclanthology.org/2023.arabicnlp-1.54.bib

https://aclanthology.org/2023.arabicnlp-1.54/

@inproceedings{qachfar-verma-2023-redaspersuasion-araieval, title = "{R}e{DASP}ersuasion at {A}r{AIE}val Shared Task: Multilingual and Monolingual Models For {A}rabic Persuasion Detection", author = "Qachfar, Fatima Zahra and Verma, Rakesh", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.54", doi = "10.18653/v1/2023.arabicnlp-1.54", pages = "549--557", abstract = "To enhance persuasion detection, we investigate the use of multilingual systems on Arabic data by conducting a total of 22 experiments using baselines, multilingual, and monolingual language transformers. Our aim is to provide a comprehensive evaluation of the various systems employed throughout this task, with the ultimate goal of comparing their performance and identifying the most effective approach. Our empirical analysis shows that *ReDASPersuasion* system performs best when combined with multilingual {``}XLM-RoBERTa{''} and monolingual pre-trained transformers on Arabic dialects like {``}CAMeLBERT-DA SA{''} depending on the NLP classification task.", }

To enhance persuasion detection, we investigate the use of multilingual systems on Arabic data by conducting a total of 22 experiments using baselines, multilingual, and monolingual language transformers. Our aim is to provide a comprehensive evaluation of the various systems employed throughout this task, with the ultimate goal of comparing their performance and identifying the most effective approach. Our empirical analysis shows that *ReDASPersuasion* system performs best when combined with multilingual {``}XLM-RoBERTa{''} and monolingual pre-trained transformers on Arabic dialects like {``}CAMeLBERT-DA SA{''} depending on the NLP classification task.

ReDASPersuasion at ArAIEval Shared Task: Multilingual and Monolingual Models For Arabic Persuasion Detection

arabicnlp-1.54

Poster

https://aclanthology.org/2023.arabicnlp-1.55.bib

https://aclanthology.org/2023.arabicnlp-1.55/

@inproceedings{lamsiyah-etal-2023-ul-um6p, title = "{UL} {\&} {UM}6{P} at {A}r{AIE}val Shared Task: Transformer-based model for Persuasion Techniques and Disinformation detection in {A}rabic", author = "Lamsiyah, Salima and El Mahdaouy, Abdelkader and Alami, Hamza and Berrada, Ismail and Schommer, Christoph", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.55", doi = "10.18653/v1/2023.arabicnlp-1.55", pages = "558--564", abstract = "In this paper, we introduce our participating system to the ArAIEval Shared Task, addressing both the detection of persuasion techniques and disinformation tasks. Our proposed system employs a pre-trained transformer-based language model for Arabic, alongside a classifier. We have assessed the performance of three Arabic Pre-trained Language Models (PLMs) for sentence encoding. Additionally, to enhance our model{'}s performance, we have explored various training objectives, including Cross-Entropy loss, regularized Mixup loss, asymmetric multi-label loss, and Focal Tversky loss. On the official test set, our system has achieved micro-F1 scores of 0.7515, 0.5666, 0.904, and 0.8333 for Sub-Task 1A, Sub-Task 1B, Sub-Task 2A, and Sub-Task 2B, respectively. Furthermore, our system has secured the 4th, 1st, 3rd, and 2nd positions, respectively, among all participating systems in sub-tasks 1A, 1B, 2A, and 2B of the ArAIEval shared task.", }

In this paper, we introduce our participating system to the ArAIEval Shared Task, addressing both the detection of persuasion techniques and disinformation tasks. Our proposed system employs a pre-trained transformer-based language model for Arabic, alongside a classifier. We have assessed the performance of three Arabic Pre-trained Language Models (PLMs) for sentence encoding. Additionally, to enhance our model{'}s performance, we have explored various training objectives, including Cross-Entropy loss, regularized Mixup loss, asymmetric multi-label loss, and Focal Tversky loss. On the official test set, our system has achieved micro-F1 scores of 0.7515, 0.5666, 0.904, and 0.8333 for Sub-Task 1A, Sub-Task 1B, Sub-Task 2A, and Sub-Task 2B, respectively. Furthermore, our system has secured the 4th, 1st, 3rd, and 2nd positions, respectively, among all participating systems in sub-tasks 1A, 1B, 2A, and 2B of the ArAIEval shared task.

UL & UM6P at ArAIEval Shared Task: Transformer-based model for Persuasion Techniques and Disinformation detection in Arabic

arabicnlp-1.55

Poster

https://aclanthology.org/2023.arabicnlp-1.56.bib

https://aclanthology.org/2023.arabicnlp-1.56/

@inproceedings{el-sayed-etal-2023-aast, title = "{AAST}-{NLP} at {A}r{AIE}val Shared Task: Tackling Persuasion technique and Disinformation Detection using Pre-Trained Language Models On Imbalanced Datasets", author = "El-Sayed, Ahmed and Nasr, Omar and Elmadany, Noureldin", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.56", doi = "10.18653/v1/2023.arabicnlp-1.56", pages = "565--569", abstract = "This paper presents the pipeline developed by the AAST-NLP team to address both the persuasion technique detection and disinformation detection shared tasks. The proposed system for all the tasks{'} sub-tasks consisted of preprocessing the data and finetuning AraBERT on the given datasets, in addition to several procedures performed for each subtask to adapt to the problems faced in it. The previously described system was used in addition to Dice loss as the loss function for sub-task 1A, which consisted of a binary classification problem. In that sub-task, the system came in eleventh place. We trained the AraBERT for task 1B, which was a multi-label problem with 24 distinct labels, using binary cross-entropy to train a classifier for each label. On that sub-task, the system came in third place. We utilised AraBERT with Dice loss on both subtasks 2A and 2B, ranking second and third among the proposed models for the respective subtasks.", }

This paper presents the pipeline developed by the AAST-NLP team to address both the persuasion technique detection and disinformation detection shared tasks. The proposed system for all the tasks{'} sub-tasks consisted of preprocessing the data and finetuning AraBERT on the given datasets, in addition to several procedures performed for each subtask to adapt to the problems faced in it. The previously described system was used in addition to Dice loss as the loss function for sub-task 1A, which consisted of a binary classification problem. In that sub-task, the system came in eleventh place. We trained the AraBERT for task 1B, which was a multi-label problem with 24 distinct labels, using binary cross-entropy to train a classifier for each label. On that sub-task, the system came in third place. We utilised AraBERT with Dice loss on both subtasks 2A and 2B, ranking second and third among the proposed models for the respective subtasks.

AAST-NLP at ArAIEval Shared Task: Tackling Persuasion technique and Disinformation Detection using Pre-Trained Language Models On Imbalanced Datasets

arabicnlp-1.56

Poster

https://aclanthology.org/2023.arabicnlp-1.57.bib

https://aclanthology.org/2023.arabicnlp-1.57/

@inproceedings{deka-revi-2023-pd, title = "{PD}-{AR} at {A}r{AIE}val Shared Task: A {BERT}-Centric Approach to Tackle {A}rabic Disinformation", author = "Deka, Pritam and Revi, Ashwathy", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.57", doi = "10.18653/v1/2023.arabicnlp-1.57", pages = "570--575", abstract = "This work explores Arabic disinformation identification, a crucial task in natural language processing, using a state-of-the-art NLP model. We highlight the performance of our system model against baseline models, including multilingual and Arabic-specific ones, and showcase the effectiveness of domain-specific pre-trained models. This work advocates for the adoption of tailored pre-trained models in NLP, emphasizing their significance in understanding diverse languages. By merging advanced NLP techniques with domain-specific pre-training, it advances Arabic disinformation identification.", }

This work explores Arabic disinformation identification, a crucial task in natural language processing, using a state-of-the-art NLP model. We highlight the performance of our system model against baseline models, including multilingual and Arabic-specific ones, and showcase the effectiveness of domain-specific pre-trained models. This work advocates for the adoption of tailored pre-trained models in NLP, emphasizing their significance in understanding diverse languages. By merging advanced NLP techniques with domain-specific pre-training, it advances Arabic disinformation identification.

PD-AR at ArAIEval Shared Task: A BERT-Centric Approach to Tackle Arabic Disinformation

arabicnlp-1.57

Poster

https://aclanthology.org/2023.arabicnlp-1.58.bib

https://aclanthology.org/2023.arabicnlp-1.58/

@inproceedings{xiao-alam-2023-nexus, title = "Nexus at {A}r{AIE}val Shared Task: Fine-Tuning {A}rabic Language Models for Propaganda and Disinformation Detection", author = "Xiao, Yunze and Alam, Firoj", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.58", doi = "10.18653/v1/2023.arabicnlp-1.58", pages = "576--582", abstract = "The spread of disinformation and propagandistic content poses a threat to societal harmony, undermining informed decision-making and trust in reliable sources. Online platforms often serve as breeding grounds for such content, and malicious actors exploit the vulnerabilities of audiences to shape public opinion. Although there have been research efforts aimed at the automatic identification of disinformation and propaganda in social media content, there remain challenges in terms of performance. The ArAIEval shared task aims to further research on these particular issues within the context of the Arabic language. In this paper, we discuss our participation in these shared tasks. We competed in subtasks 1A and 2A, where our submitted system secured positions 9th and 10th, respectively. Our experiments consist of fine-tuning transformer models and using zero- and few-shot learning with GPT-4.", }

The spread of disinformation and propagandistic content poses a threat to societal harmony, undermining informed decision-making and trust in reliable sources. Online platforms often serve as breeding grounds for such content, and malicious actors exploit the vulnerabilities of audiences to shape public opinion. Although there have been research efforts aimed at the automatic identification of disinformation and propaganda in social media content, there remain challenges in terms of performance. The ArAIEval shared task aims to further research on these particular issues within the context of the Arabic language. In this paper, we discuss our participation in these shared tasks. We competed in subtasks 1A and 2A, where our submitted system secured positions 9th and 10th, respectively. Our experiments consist of fine-tuning transformer models and using zero- and few-shot learning with GPT-4.

Nexus at ArAIEval Shared Task: Fine-Tuning Arabic Language Models for Propaganda and Disinformation Detection

arabicnlp-1.58

2311.03184

Poster

https://aclanthology.org/2023.arabicnlp-1.59.bib

https://aclanthology.org/2023.arabicnlp-1.59/

@inproceedings{azizov-etal-2023-frank, title = "Frank at {A}r{AIE}val Shared Task: {A}rabic Persuasion and Disinformation: The Power of Pretrained Models", author = "Azizov, Dilshod and Li, Jiyong and Liang, Shangsong", editor = "Sawaf, Hassan and El-Beltagy, Samhaa and Zaghouani, Wajdi and Magdy, Walid and Abdelali, Ahmed and Tomeh, Nadi and Abu Farha, Ibrahim and Habash, Nizar and Khalifa, Salam and Keleg, Amr and Haddad, Hatem and Zitouni, Imed and Mrini, Khalil and Almatham, Rawan", booktitle = "Proceedings of ArabicNLP 2023", month = dec, year = "2023", address = "Singapore (Hybrid)", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.arabicnlp-1.59", doi = "10.18653/v1/2023.arabicnlp-1.59", pages = "583--588", abstract = {In this work, we present our systems developed for {``}ArAIEval{''} shared task of ArabicNLP 2023 (CITATION). We used an mBERT transformer for Subtask 1A, which targets persuasion in Arabic tweets, and we used the MARBERT transformer for Subtask 2A to identify disinformation in Arabic tweets. Our persuasion detection system achieved micro-F1 of \textbf{0.745} by surpassing the baseline by 13.2{\%}, and registered a macro-F1 of 0.717 based on leaderboard scores. Similarly, our disinformation system recorded a micro-F1 of \textbf{0.816}, besting the na{\"\i}ve majority by 6.7{\%}, with a macro-F1 of 0.637. Furthermore, we present our preliminary results on a variety of pre-trained models. In terms of overall ranking, our systems placed $7^\text{th}$ out of 16 and $12^\text{th}$ out of 17 teams for Subtasks 1A and 2A, respectively.}, }

In this work, we present our systems developed for {``}ArAIEval{''} shared task of ArabicNLP 2023 (CITATION). We used an mBERT transformer for Subtask 1A, which targets persuasion in Arabic tweets, and we used the MARBERT transformer for Subtask 2A to identify disinformation in Arabic tweets. Our persuasion detection system achieved micro-F1 of \textbf{0.745} by surpassing the baseline by 13.2{\%}, and registered a macro-F1 of 0.717 based on leaderboard scores. Similarly, our disinformation system recorded a micro-F1 of \textbf{0.816}, besting the na{\"\i}ve majority by 6.7{\%}, with a macro-F1 of 0.637. Furthermore, we present our preliminary results on a variety of pre-trained models. In terms of overall ranking, our systems placed $7^\text{th}$ out of 16 and $12^\text{th}$ out of 17 teams for Subtasks 1A and 2A, respectively.

Frank at ArAIEval Shared Task: Arabic Persuasion and Disinformation: The Power of Pretrained Models

arabicnlp-1.59

Poster