Datasets:

EMNLP

/

EMNLP2023-papers

like 0

Follow

Conference on Empirical Methods in Natural Language Processing 1

https://aclanthology.org/2023.findings-emnlp.920.bib

https://aclanthology.org/2023.findings-emnlp.920/

@inproceedings{liu-etal-2023-enhancing-scalability, title = "Enhancing Scalability of Pre-trained Language Models via Efficient Parameter Sharing", author = "Liu, Peiyu and Gao, Ze-Feng and Chen, Yushuo and Zhao, Xin 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.920", doi = "10.18653/v1/2023.findings-emnlp.920", pages = "13771--13785", abstract = "In this paper, we propose a highly parameter-efficient approach to scaling pre-trained language models (PLMs) to a deeper model depth. Unlike prior work that shares all parameters or uses extra blocks, we design a more capable parameter-sharing architecture based on matrix product operator (MPO), an efficient tensor decomposition method to factorize the parameter matrix into a set of local tensors. Based on such a decomposition, we share the important local tensor across all layers for reducing the model size and meanwhile keep layer-specific tensors (also using Adapters) for enhancing the adaptation flexibility. To improve the model training, we further propose a stable initialization algorithm tailored for the MPO-based architecture. Extensive experiments have demonstrated the effectiveness of our proposed model in enhancing scalability and achieving higher performance (i.e., with fewer parameters than BERT-base, we successfully scale the model depth by a factor of 4x and even achieve 0.1 points higher than BERT-large for GLUE score). The code to reproduce the results of this paper can be found at https://github.com/RUCAIBox/MPOBERT-code.", }

In this paper, we propose a highly parameter-efficient approach to scaling pre-trained language models (PLMs) to a deeper model depth. Unlike prior work that shares all parameters or uses extra blocks, we design a more capable parameter-sharing architecture based on matrix product operator (MPO), an efficient tensor decomposition method to factorize the parameter matrix into a set of local tensors. Based on such a decomposition, we share the important local tensor across all layers for reducing the model size and meanwhile keep layer-specific tensors (also using Adapters) for enhancing the adaptation flexibility. To improve the model training, we further propose a stable initialization algorithm tailored for the MPO-based architecture. Extensive experiments have demonstrated the effectiveness of our proposed model in enhancing scalability and achieving higher performance (i.e., with fewer parameters than BERT-base, we successfully scale the model depth by a factor of 4x and even achieve 0.1 points higher than BERT-large for GLUE score). The code to reproduce the results of this paper can be found at https://github.com/RUCAIBox/MPOBERT-code.

Enhancing Scalability of Pre-trained Language Models via Efficient Parameter Sharing

findings-emnlp.920

Poster

https://aclanthology.org/2023.findings-emnlp.921.bib

https://aclanthology.org/2023.findings-emnlp.921/

@inproceedings{kho-etal-2023-boosting, title = "Boosting Prompt-Based Self-Training With Mapping-Free Automatic Verbalizer for Multi-Class Classification", author = "Kho, Yookyung and Kim, Jaehee and Kang, Pilsung", 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.921", doi = "10.18653/v1/2023.findings-emnlp.921", pages = "13786--13800", abstract = "Recently, prompt-based fine-tuning has garnered considerable interest as a core technique for few-shot text classification task. This approach reformulates the fine-tuning objective to align with the Masked Language Modeling (MLM) objective. Leveraging unlabeled data, prompt-based self-training has shown greater effectiveness in binary and three-class classification. However, prompt-based self-training for multi-class classification has not been adequately investigated, despite its significant applicability to real-world scenarios. Moreover, extending current methods to multi-class classification suffers from the verbalizer that extracts the predicted value of manually pre-defined single label word for each class from MLM predictions. Consequently, we introduce a novel, efficient verbalizer structure, named Mapping-free Automatic Verbalizer (MAV). Comprising two fully connected layers, MAV serves as a trainable verbalizer that automatically extracts the requisite word features for classification by capitalizing on all available information from MLM predictions. Experimental results on five multi-class classification datasets indicate MAV{'}s superior self-training efficacy.", }

Recently, prompt-based fine-tuning has garnered considerable interest as a core technique for few-shot text classification task. This approach reformulates the fine-tuning objective to align with the Masked Language Modeling (MLM) objective. Leveraging unlabeled data, prompt-based self-training has shown greater effectiveness in binary and three-class classification. However, prompt-based self-training for multi-class classification has not been adequately investigated, despite its significant applicability to real-world scenarios. Moreover, extending current methods to multi-class classification suffers from the verbalizer that extracts the predicted value of manually pre-defined single label word for each class from MLM predictions. Consequently, we introduce a novel, efficient verbalizer structure, named Mapping-free Automatic Verbalizer (MAV). Comprising two fully connected layers, MAV serves as a trainable verbalizer that automatically extracts the requisite word features for classification by capitalizing on all available information from MLM predictions. Experimental results on five multi-class classification datasets indicate MAV{'}s superior self-training efficacy.

Boosting Prompt-Based Self-Training With Mapping-Free Automatic Verbalizer for Multi-Class Classification

findings-emnlp.921

2312.04982

Poster

https://aclanthology.org/2023.findings-emnlp.922.bib

https://aclanthology.org/2023.findings-emnlp.922/

@inproceedings{dada-etal-2023-impact, title = "On the Impact of Cross-Domain Data on {G}erman Language Models", author = "Dada, Amin and Chen, Aokun and Peng, Cheng and Smith, Kaleb and Idrissi-Yaghir, Ahmad and Seibold, Constantin and Li, Jianning and Heiliger, Lars and Friedrich, Christoph and Truhn, Daniel and Egger, Jan and Bian, Jiang and Kleesiek, Jens and Wu, Yonghui", 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.922", doi = "10.18653/v1/2023.findings-emnlp.922", pages = "13801--13813", abstract = "Traditionally, large language models have been either trained on general web crawls or domain-specific data. However, recent successes of generative large language models, have shed light on the benefits of cross-domain datasets. To examine the significance of prioritizing data diversity over quality, we present a German dataset comprising texts from five domains, along with another dataset aimed at containing high-quality data. Through training a series of models ranging between 122M and 750M parameters on both datasets, we conduct a comprehensive benchmark on multiple downstream tasks. Our findings demonstrate that the models trained on the cross-domain dataset outperform those trained on quality data alone, leading to improvements up to 4.45{\%} over the previous state-of-the-art.", }

Traditionally, large language models have been either trained on general web crawls or domain-specific data. However, recent successes of generative large language models, have shed light on the benefits of cross-domain datasets. To examine the significance of prioritizing data diversity over quality, we present a German dataset comprising texts from five domains, along with another dataset aimed at containing high-quality data. Through training a series of models ranging between 122M and 750M parameters on both datasets, we conduct a comprehensive benchmark on multiple downstream tasks. Our findings demonstrate that the models trained on the cross-domain dataset outperform those trained on quality data alone, leading to improvements up to 4.45{\%} over the previous state-of-the-art.

On the Impact of Cross-Domain Data on German Language Models

findings-emnlp.922

2310.07321

Poster

https://aclanthology.org/2023.findings-emnlp.923.bib

https://aclanthology.org/2023.findings-emnlp.923/

@inproceedings{kuparinen-etal-2023-dialect, title = "Dialect-to-Standard Normalization: A Large-Scale Multilingual Evaluation", author = "Kuparinen, Olli and Mileti{\'c}, Aleksandra and Scherrer, Yves", 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.923", doi = "10.18653/v1/2023.findings-emnlp.923", pages = "13814--13828", abstract = "Text normalization methods have been commonly applied to historical language or user-generated content, but less often to dialectal transcriptions. In this paper, we introduce dialect-to-standard normalization {--} i.e., mapping phonetic transcriptions from different dialects to the orthographic norm of the standard variety {--} as a distinct sentence-level character transduction task and provide a large-scale analysis of dialect-to-standard normalization methods. To this end, we compile a multilingual dataset covering four languages: Finnish, Norwegian, Swiss German and Slovene. For the two biggest corpora, we provide three different data splits corresponding to different use cases for automatic normalization. We evaluate the most successful sequence-to-sequence model architectures proposed for text normalization tasks using different tokenization approaches and context sizes. We find that a character-level Transformer trained on sliding windows of three words works best for Finnish, Swiss German and Slovene, whereas the pre-trained byT5 model using full sentences obtains the best results for Norwegian. Finally, we perform an error analysis to evaluate the effect of different data splits on model performance.", }

Text normalization methods have been commonly applied to historical language or user-generated content, but less often to dialectal transcriptions. In this paper, we introduce dialect-to-standard normalization {--} i.e., mapping phonetic transcriptions from different dialects to the orthographic norm of the standard variety {--} as a distinct sentence-level character transduction task and provide a large-scale analysis of dialect-to-standard normalization methods. To this end, we compile a multilingual dataset covering four languages: Finnish, Norwegian, Swiss German and Slovene. For the two biggest corpora, we provide three different data splits corresponding to different use cases for automatic normalization. We evaluate the most successful sequence-to-sequence model architectures proposed for text normalization tasks using different tokenization approaches and context sizes. We find that a character-level Transformer trained on sliding windows of three words works best for Finnish, Swiss German and Slovene, whereas the pre-trained byT5 model using full sentences obtains the best results for Norwegian. Finally, we perform an error analysis to evaluate the effect of different data splits on model performance.

Dialect-to-Standard Normalization: A Large-Scale Multilingual Evaluation

findings-emnlp.923

Poster

https://aclanthology.org/2023.findings-emnlp.924.bib

https://aclanthology.org/2023.findings-emnlp.924/

@inproceedings{ernst-etal-2023-examining, title = "Re-Examining Summarization Evaluation across Multiple Quality Criteria", author = "Ernst, Ori and Shapira, Ori and Dagan, Ido and Levy, Ran", 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.924", doi = "10.18653/v1/2023.findings-emnlp.924", pages = "13829--13838", abstract = "The common practice for assessing automatic evaluation metrics is to measure the correlation between their induced system rankings and those obtained by reliable human evaluation, where a higher correlation indicates a better metric. Yet, an intricate setting arises when an NLP task is evaluated by multiple Quality Criteria (QCs), like for text summarization where prominent criteria including relevance, consistency, fluency and coherence. In this paper, we challenge the soundness of this methodology when multiple QCs are involved, concretely for the summarization case. First, we show that the allegedly best metrics for certain QCs actually do not perform well, failing to detect even drastic summary corruptions with respect to the considered QC. To explain this, we show that some of the high correlations obtained in the multi-QC setup are spurious. Finally, we propose a procedure that may help detecting this effect. Overall, our findings highlight the need for further investigating metric evaluation methodologies for the multiple-QC case.", }

The common practice for assessing automatic evaluation metrics is to measure the correlation between their induced system rankings and those obtained by reliable human evaluation, where a higher correlation indicates a better metric. Yet, an intricate setting arises when an NLP task is evaluated by multiple Quality Criteria (QCs), like for text summarization where prominent criteria including relevance, consistency, fluency and coherence. In this paper, we challenge the soundness of this methodology when multiple QCs are involved, concretely for the summarization case. First, we show that the allegedly best metrics for certain QCs actually do not perform well, failing to detect even drastic summary corruptions with respect to the considered QC. To explain this, we show that some of the high correlations obtained in the multi-QC setup are spurious. Finally, we propose a procedure that may help detecting this effect. Overall, our findings highlight the need for further investigating metric evaluation methodologies for the multiple-QC case.

Re-Examining Summarization Evaluation across Multiple Quality Criteria

findings-emnlp.924

Poster

https://aclanthology.org/2023.findings-emnlp.925.bib

https://aclanthology.org/2023.findings-emnlp.925/

@inproceedings{le-luu-2023-parallel, title = "A Parallel Corpus for {V}ietnamese Central-Northern Dialect Text Transfer", author = "Le, Thang and Luu, Anh", 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.925", doi = "10.18653/v1/2023.findings-emnlp.925", pages = "13839--13855", abstract = "The Vietnamese language embodies dialectal variants closely attached to the nation{'}s three macro-regions: the Northern, Central and Southern regions. As the northern dialect forms the basis of the standard language, it{'}s considered the prestige dialect. While the northern dialect differs from the remaining two in certain aspects, it almost shares an identical lexicon with the southern dialect, making the textual attributes nearly interchangeable. In contrast, the central dialect possesses a number of unique vocabularies and is less mutually intelligible to the standard dialect. Through preliminary experiments, we observe that current NLP models do not possess understandings of the Vietnamese central dialect text, which most likely originates from the lack of resources. To facilitate research on this domain, we introduce a new parallel corpus for Vietnamese central-northern dialect text transfer. Via exhaustive benchmarking, we discover monolingual language models{'} superiority over their multilingual counterparts on the dialect transfer task. We further demonstrate that fine-tuned transfer models can seamlessly improve the performance of existing NLP systems on the central dialect domain with dedicated results in translation and text-image retrieval tasks.", }

The Vietnamese language embodies dialectal variants closely attached to the nation{'}s three macro-regions: the Northern, Central and Southern regions. As the northern dialect forms the basis of the standard language, it{'}s considered the prestige dialect. While the northern dialect differs from the remaining two in certain aspects, it almost shares an identical lexicon with the southern dialect, making the textual attributes nearly interchangeable. In contrast, the central dialect possesses a number of unique vocabularies and is less mutually intelligible to the standard dialect. Through preliminary experiments, we observe that current NLP models do not possess understandings of the Vietnamese central dialect text, which most likely originates from the lack of resources. To facilitate research on this domain, we introduce a new parallel corpus for Vietnamese central-northern dialect text transfer. Via exhaustive benchmarking, we discover monolingual language models{'} superiority over their multilingual counterparts on the dialect transfer task. We further demonstrate that fine-tuned transfer models can seamlessly improve the performance of existing NLP systems on the central dialect domain with dedicated results in translation and text-image retrieval tasks.

A Parallel Corpus for Vietnamese Central-Northern Dialect Text Transfer

findings-emnlp.925

Poster

https://aclanthology.org/2023.findings-emnlp.926.bib

https://aclanthology.org/2023.findings-emnlp.926/

@inproceedings{jacovi-etal-2023-comprehensive, title = "A Comprehensive Evaluation of Tool-Assisted Generation Strategies", author = "Jacovi, Alon and Caciularu, Avi and Herzig, Jonathan and Aharoni, Roee and Bohnet, Bernd and Geva, Mor", 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.926", doi = "10.18653/v1/2023.findings-emnlp.926", pages = "13856--13878", abstract = "A growing area of research investigates augmenting language models with tools (e.g., search engines, calculators) to overcome their shortcomings (e.g., missing or incorrect knowledge, incorrect logical inferences). Various few-shot tool-usage strategies have been proposed. However, there is no systematic and fair comparison across different strategies, or between these strategies and strong baselines that do not leverage tools. We conduct an extensive empirical analysis, finding that (1) across various datasets, example difficulty levels, and models, strong no-tool baselines are competitive to tool-assisted strategies, implying that effectively using tools with in-context demonstrations is a difficult unsolved problem; (2) for knowledge-retrieval tasks, strategies that *refine* incorrect outputs with tools outperform strategies that retrieve relevant information *ahead of* or *during generation*; (3) tool-assisted strategies are expensive in the number of tokens they require to work{---}incurring additional costs by orders of magnitude{---}which does not translate into significant improvement in performance. Overall, our findings suggest that few-shot tool integration is still an open challenge, emphasizing the need for comprehensive evaluations of future strategies to accurately assess their *benefits* and *costs*.", }

A growing area of research investigates augmenting language models with tools (e.g., search engines, calculators) to overcome their shortcomings (e.g., missing or incorrect knowledge, incorrect logical inferences). Various few-shot tool-usage strategies have been proposed. However, there is no systematic and fair comparison across different strategies, or between these strategies and strong baselines that do not leverage tools. We conduct an extensive empirical analysis, finding that (1) across various datasets, example difficulty levels, and models, strong no-tool baselines are competitive to tool-assisted strategies, implying that effectively using tools with in-context demonstrations is a difficult unsolved problem; (2) for knowledge-retrieval tasks, strategies that *refine* incorrect outputs with tools outperform strategies that retrieve relevant information *ahead of* or *during generation*; (3) tool-assisted strategies are expensive in the number of tokens they require to work{---}incurring additional costs by orders of magnitude{---}which does not translate into significant improvement in performance. Overall, our findings suggest that few-shot tool integration is still an open challenge, emphasizing the need for comprehensive evaluations of future strategies to accurately assess their *benefits* and *costs*.

A Comprehensive Evaluation of Tool-Assisted Generation Strategies

findings-emnlp.926

2310.10062

Poster

https://aclanthology.org/2023.findings-emnlp.927.bib

https://aclanthology.org/2023.findings-emnlp.927/

@inproceedings{xu-etal-2023-inheritsumm, title = "{I}nherit{S}umm: A General, Versatile and Compact Summarizer by Distilling from {GPT}", author = "Xu, Yichong and Xu, Ruochen and Iter, Dan and Liu, Yang and Wang, Shuohang and Zhu, Chenguang and Zeng, Michael", 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.927", doi = "10.18653/v1/2023.findings-emnlp.927", pages = "13879--13892", abstract = "While large models such as GPT-3 demonstrate exceptional performance in zeroshot and fewshot summarization tasks, their extensive serving and fine-tuning costs hinder their utilization in various applications. Conversely, previous studies have found that although automatic metrics tend to favor smaller fine-tuned models, the quality of the summaries they generate is inferior to that of larger models like GPT-3 when assessed by human evaluators. To address this issue, we propose InheritSumm, a versatile and compact summarization model derived from GPT-3.5 through distillation. InheritSumm not only exhibits comparable zeroshot and fewshot summarization capabilities to GPT-3.5 but is also sufficiently compact for fine-tuning purposes. Experimental results demonstrate that InheritSumm achieves similar or superior performance to GPT-3.5 in zeroshot and fewshot settings. Furthermore, it outperforms the previously established best small models in both prefix-tuning and full-data fine-tuning scenarios.", }

While large models such as GPT-3 demonstrate exceptional performance in zeroshot and fewshot summarization tasks, their extensive serving and fine-tuning costs hinder their utilization in various applications. Conversely, previous studies have found that although automatic metrics tend to favor smaller fine-tuned models, the quality of the summaries they generate is inferior to that of larger models like GPT-3 when assessed by human evaluators. To address this issue, we propose InheritSumm, a versatile and compact summarization model derived from GPT-3.5 through distillation. InheritSumm not only exhibits comparable zeroshot and fewshot summarization capabilities to GPT-3.5 but is also sufficiently compact for fine-tuning purposes. Experimental results demonstrate that InheritSumm achieves similar or superior performance to GPT-3.5 in zeroshot and fewshot settings. Furthermore, it outperforms the previously established best small models in both prefix-tuning and full-data fine-tuning scenarios.

InheritSumm: A General, Versatile and Compact Summarizer by Distilling from GPT

findings-emnlp.927

2305.13083

Poster

https://aclanthology.org/2023.findings-emnlp.928.bib

https://aclanthology.org/2023.findings-emnlp.928/

@inproceedings{ilagan-2023-learning, title = "Learning to love diligent trolls: Accounting for rater effects in the dialogue safety task", author = "Ilagan, Michael", 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.928", doi = "10.18653/v1/2023.findings-emnlp.928", pages = "13893--13899", abstract = "Chatbots have the risk of generating offensive utterances, which must be avoided. Post-deployment, one way for a chatbot to continuously improve is to source utterance/label pairs from feedback by live users. However, among users are trolls, who provide training examples with incorrect labels. To de-troll training data, previous work removed training examples that have high user-aggregated cross-validation (CV) error. However, CV is expensive; and in a coordinated attack, CV may be overwhelmed by trolls in number and in consistency among themselves. In the present work, I address both limitations by proposing a solution inspired by methodology in automated essay scoring (AES): have multiple users rate each utterance, then perform latent class analysis (LCA) to infer correct labels. As it does not require GPU computations, LCA is inexpensive. In experiments, I found that the AES-like solution can infer training labels with high accuracy when trolls are consistent, even when trolls are the majority.", }

Chatbots have the risk of generating offensive utterances, which must be avoided. Post-deployment, one way for a chatbot to continuously improve is to source utterance/label pairs from feedback by live users. However, among users are trolls, who provide training examples with incorrect labels. To de-troll training data, previous work removed training examples that have high user-aggregated cross-validation (CV) error. However, CV is expensive; and in a coordinated attack, CV may be overwhelmed by trolls in number and in consistency among themselves. In the present work, I address both limitations by proposing a solution inspired by methodology in automated essay scoring (AES): have multiple users rate each utterance, then perform latent class analysis (LCA) to infer correct labels. As it does not require GPU computations, LCA is inexpensive. In experiments, I found that the AES-like solution can infer training labels with high accuracy when trolls are consistent, even when trolls are the majority.

Learning to love diligent trolls: Accounting for rater effects in the dialogue safety task

findings-emnlp.928

2310.19271

Poster

https://aclanthology.org/2023.findings-emnlp.929.bib

https://aclanthology.org/2023.findings-emnlp.929/

@inproceedings{kang-etal-2023-chatgpt, title = "Can {C}hat{GPT} Perform Reasoning Using the {IRAC} Method in Analyzing Legal Scenarios Like a Lawyer?", author = "Kang, Xiaoxi and Qu, Lizhen and Soon, Lay-Ki and Trakic, Adnan and Zhuo, Terry and Emerton, Patrick and Grant, Genevieve", 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.929", doi = "10.18653/v1/2023.findings-emnlp.929", pages = "13900--13923", abstract = "Large Language Models (LLMs), such as ChatGPT, have drawn a lot of attentions recently in the legal domain due to its emergent ability to tackle a variety of legal tasks. However, it is still unknown if LLMs are able to analyze a legal case and perform reasoning in the same manner as lawyers. Therefore, we constructed a novel corpus consisting of scenarios pertain to Contract Acts Malaysia and Australian Social Act for Dependent Child. ChatGPT is applied to perform analysis on the corpus using the IRAC method, which is a framework widely used by legal professionals for organizing legal analysis. Each scenario in the corpus is annotated with a complete IRAC analysis in a semi-structured format so that both machines and legal professionals are able to interpret and understand the annotations. In addition, we conducted the first empirical assessment of ChatGPT for IRAC analysis in order to understand how well it aligns with the analysis of legal professionals. Our experimental results shed lights on possible future research directions to improve alignments between LLMs and legal experts in terms of legal reasoning.", }

Large Language Models (LLMs), such as ChatGPT, have drawn a lot of attentions recently in the legal domain due to its emergent ability to tackle a variety of legal tasks. However, it is still unknown if LLMs are able to analyze a legal case and perform reasoning in the same manner as lawyers. Therefore, we constructed a novel corpus consisting of scenarios pertain to Contract Acts Malaysia and Australian Social Act for Dependent Child. ChatGPT is applied to perform analysis on the corpus using the IRAC method, which is a framework widely used by legal professionals for organizing legal analysis. Each scenario in the corpus is annotated with a complete IRAC analysis in a semi-structured format so that both machines and legal professionals are able to interpret and understand the annotations. In addition, we conducted the first empirical assessment of ChatGPT for IRAC analysis in order to understand how well it aligns with the analysis of legal professionals. Our experimental results shed lights on possible future research directions to improve alignments between LLMs and legal experts in terms of legal reasoning.

Can ChatGPT Perform Reasoning Using the IRAC Method in Analyzing Legal Scenarios Like a Lawyer?

findings-emnlp.929

2310.14880

Poster

https://aclanthology.org/2023.findings-emnlp.930.bib

https://aclanthology.org/2023.findings-emnlp.930/

@inproceedings{gupta-etal-2023-coverage, title = "Coverage-based Example Selection for In-Context Learning", author = "Gupta, Shivanshu and Gardner, Matt and Singh, Sameer", 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.930", doi = "10.18653/v1/2023.findings-emnlp.930", pages = "13924--13950", abstract = "In-context learning (ICL), the ability of large language models to perform novel tasks by conditioning on a prompt with a few task examples, requires these examples to be informative about the test instance. The standard approach of independently ranking and selecting the most similar examples selects redundant examples while omitting important information. In this work, we show that BERTScore-Recall (BSR) selects better examples that demonstrate more of the salient aspects, e.g. reasoning patterns, of the test input. We further extend BSR and many standard metrics to easily optimizable set-level metrics, giving still better coverage of those salient aspects. On 15 datasets spanning 6 tasks and with 7 diverse LLMs, we show that (1) BSR is the superior metric for in-context example selection across the board, and (2) for compositional tasks, set selection using Set-BSR outperforms independent ranking by up to 17 points on average and, despite being training-free, surpasses methods that leverage task or LLM-specific training.", }

In-context learning (ICL), the ability of large language models to perform novel tasks by conditioning on a prompt with a few task examples, requires these examples to be informative about the test instance. The standard approach of independently ranking and selecting the most similar examples selects redundant examples while omitting important information. In this work, we show that BERTScore-Recall (BSR) selects better examples that demonstrate more of the salient aspects, e.g. reasoning patterns, of the test input. We further extend BSR and many standard metrics to easily optimizable set-level metrics, giving still better coverage of those salient aspects. On 15 datasets spanning 6 tasks and with 7 diverse LLMs, we show that (1) BSR is the superior metric for in-context example selection across the board, and (2) for compositional tasks, set selection using Set-BSR outperforms independent ranking by up to 17 points on average and, despite being training-free, surpasses methods that leverage task or LLM-specific training.

Coverage-based Example Selection for In-Context Learning

findings-emnlp.930

2305.14907

Poster

https://aclanthology.org/2023.findings-emnlp.931.bib

https://aclanthology.org/2023.findings-emnlp.931/

@inproceedings{xu-etal-2023-structural, title = "Are Structural Concepts Universal in Transformer Language Models? Towards Interpretable Cross-Lingual Generalization", author = "Xu, Ningyu and Zhang, Qi and Ye, Jingting and Zhang, Menghan and Huang, Xuanjing", 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.931", doi = "10.18653/v1/2023.findings-emnlp.931", pages = "13951--13976", abstract = "Large language models (LLMs) have exhibited considerable cross-lingual generalization abilities, whereby they implicitly transfer knowledge across languages. However, the transfer is not equally successful for all languages, especially for low-resource ones, which poses an ongoing challenge. It is unclear whether we have reached the limits of implicit cross-lingual generalization and if explicit knowledge transfer is viable. In this paper, we investigate the potential for explicitly aligning conceptual correspondence between languages to enhance cross-lingual generalization. Using the syntactic aspect of language as a testbed, our analyses of 43 languages reveal a high degree of alignability among the spaces of structural concepts within each language for both encoder-only and decoder-only LLMs. We then propose a meta-learning-based method to learn to align conceptual spaces of different languages, which facilitates zero-shot and few-shot generalization in concept classification and also offers insights into the cross-lingual in-context learning phenomenon. Experiments on syntactic analysis tasks show that our approach achieves competitive results with state-of-the-art methods and narrows the performance gap between languages, particularly benefiting those with limited resources.", }

Large language models (LLMs) have exhibited considerable cross-lingual generalization abilities, whereby they implicitly transfer knowledge across languages. However, the transfer is not equally successful for all languages, especially for low-resource ones, which poses an ongoing challenge. It is unclear whether we have reached the limits of implicit cross-lingual generalization and if explicit knowledge transfer is viable. In this paper, we investigate the potential for explicitly aligning conceptual correspondence between languages to enhance cross-lingual generalization. Using the syntactic aspect of language as a testbed, our analyses of 43 languages reveal a high degree of alignability among the spaces of structural concepts within each language for both encoder-only and decoder-only LLMs. We then propose a meta-learning-based method to learn to align conceptual spaces of different languages, which facilitates zero-shot and few-shot generalization in concept classification and also offers insights into the cross-lingual in-context learning phenomenon. Experiments on syntactic analysis tasks show that our approach achieves competitive results with state-of-the-art methods and narrows the performance gap between languages, particularly benefiting those with limited resources.

Are Structural Concepts Universal in Transformer Language Models? Towards Interpretable Cross-Lingual Generalization

findings-emnlp.931

2310.12794

Poster

https://aclanthology.org/2023.findings-emnlp.932.bib

https://aclanthology.org/2023.findings-emnlp.932/

@inproceedings{kaffee-etal-2023-thorny, title = "Thorny Roses: Investigating the Dual Use Dilemma in Natural Language Processing", author = "Kaffee, Lucie-Aim{\'e}e and Arora, Arnav and Talat, Zeerak and Augenstein, Isabelle", 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.932", doi = "10.18653/v1/2023.findings-emnlp.932", pages = "13977--13998", abstract = "Dual use, the intentional, harmful reuse of technology and scientific artefacts, is an ill-defined problem within the context of Natural Language Processing (NLP). As large language models (LLMs) have advanced in their capabilities and become more accessible, the risk of their intentional misuse becomes more prevalent. To prevent such intentional malicious use, it is necessary for NLP researchers and practitioners to understand and mitigate the risks of their research. Hence, we present an NLP-specific definition of dual use informed by researchers and practitioners in the field. Further, we propose a checklist focusing on dual-use in NLP, that can be integrated into existing conference ethics-frameworks. The definition and checklist are created based on a survey of NLP researchers and practitioners.", }

Dual use, the intentional, harmful reuse of technology and scientific artefacts, is an ill-defined problem within the context of Natural Language Processing (NLP). As large language models (LLMs) have advanced in their capabilities and become more accessible, the risk of their intentional misuse becomes more prevalent. To prevent such intentional malicious use, it is necessary for NLP researchers and practitioners to understand and mitigate the risks of their research. Hence, we present an NLP-specific definition of dual use informed by researchers and practitioners in the field. Further, we propose a checklist focusing on dual-use in NLP, that can be integrated into existing conference ethics-frameworks. The definition and checklist are created based on a survey of NLP researchers and practitioners.

Thorny Roses: Investigating the Dual Use Dilemma in Natural Language Processing

findings-emnlp.932

2304.08315

Poster

https://aclanthology.org/2023.findings-emnlp.933.bib

https://aclanthology.org/2023.findings-emnlp.933/

@inproceedings{bohra-etal-2023-byoc, title = "{BYOC}: Personalized Few-Shot Classification with Co-Authored Class Descriptions", author = "Bohra, Arth and Verkes, Govert and Harutyunyan, Artem and Weinberger, Pascal and Campagna, Giovanni", 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.933", doi = "10.18653/v1/2023.findings-emnlp.933", pages = "13999--14015", abstract = "Text classification is a well-studied and versatile building block for many NLP applications. Yet, existing approaches require either large annotated corpora to train a model with or, when using large language models as a base, require carefully crafting the prompt as well as using a long context that can fit many examples. As a result, it is not possible for end-users to build classifiers for themselves. To address this issue, we propose a novel approach to few-shot text classification using an LLM. Rather than few-shot examples, the LLM is prompted with descriptions of the salient features of each class. These descriptions are coauthored by the user and the LLM interactively: while the user annotates each few-shot example, the LLM asks relevant questions that the user answers. Examples, questions, and answers are summarized to form the classification prompt. Our experiments show that our approach yields high accuracy classifiers, within 79{\%} of the performance of models trained with significantly larger datasets while using only 1{\%} of their training sets. Additionally, in a study with 30 participants, we show that end-users are able to build classifiers to suit their specific needs. The personalized classifiers show an average accuracy of 90{\%}, which is 15{\%} higher than the state-of-the-art approach.", }

Text classification is a well-studied and versatile building block for many NLP applications. Yet, existing approaches require either large annotated corpora to train a model with or, when using large language models as a base, require carefully crafting the prompt as well as using a long context that can fit many examples. As a result, it is not possible for end-users to build classifiers for themselves. To address this issue, we propose a novel approach to few-shot text classification using an LLM. Rather than few-shot examples, the LLM is prompted with descriptions of the salient features of each class. These descriptions are coauthored by the user and the LLM interactively: while the user annotates each few-shot example, the LLM asks relevant questions that the user answers. Examples, questions, and answers are summarized to form the classification prompt. Our experiments show that our approach yields high accuracy classifiers, within 79{\%} of the performance of models trained with significantly larger datasets while using only 1{\%} of their training sets. Additionally, in a study with 30 participants, we show that end-users are able to build classifiers to suit their specific needs. The personalized classifiers show an average accuracy of 90{\%}, which is 15{\%} higher than the state-of-the-art approach.

BYOC: Personalized Few-Shot Classification with Co-Authored Class Descriptions

findings-emnlp.933

2310.06111

Poster

https://aclanthology.org/2023.findings-emnlp.934.bib

https://aclanthology.org/2023.findings-emnlp.934/

@inproceedings{khalighinejad-etal-2023-approximating, title = "Approximating {CKY} with Transformers", author = "Khalighinejad, Ghazal and Liu, Ollie and Wiseman, Sam", 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.934", doi = "10.18653/v1/2023.findings-emnlp.934", pages = "14016--14030", abstract = "We investigate the ability of transformer models to approximate the CKY algorithm, using them to directly predict a sentence{'}s parse and thus avoid the CKY algorithm{'}s cubic dependence on sentence length. We find that on standard constituency parsing benchmarks this approach achieves competitive or better performance than comparable parsers that make use of CKY, while being faster. We also evaluate the viability of this approach for parsing under \textit{random} PCFGs. Here we find that performance declines as the grammar becomes more ambiguous, suggesting that the transformer is not fully capturing the CKY computation. However, we also find that incorporating additional inductive bias is helpful, and we propose a novel approach that makes use of gradients with respect to chart representations in predicting the parse, in analogy with the CKY algorithm being a subgradient of a partition function variant with respect to the chart.", }

We investigate the ability of transformer models to approximate the CKY algorithm, using them to directly predict a sentence{'}s parse and thus avoid the CKY algorithm{'}s cubic dependence on sentence length. We find that on standard constituency parsing benchmarks this approach achieves competitive or better performance than comparable parsers that make use of CKY, while being faster. We also evaluate the viability of this approach for parsing under \textit{random} PCFGs. Here we find that performance declines as the grammar becomes more ambiguous, suggesting that the transformer is not fully capturing the CKY computation. However, we also find that incorporating additional inductive bias is helpful, and we propose a novel approach that makes use of gradients with respect to chart representations in predicting the parse, in analogy with the CKY algorithm being a subgradient of a partition function variant with respect to the chart.

Approximating CKY with Transformers

findings-emnlp.934

2305.02386

Poster

https://aclanthology.org/2023.findings-emnlp.935.bib

https://aclanthology.org/2023.findings-emnlp.935/

@inproceedings{gupta-etal-2023-dialguide, title = "{D}ial{G}uide: Aligning Dialogue Model Behavior with Developer Guidelines", author = "Gupta, Prakhar and Liu, Yang and Jin, Di and Hedayatnia, Behnam and Gella, Spandana and Liu, Sijia and Lange, Patrick and Hirschberg, Julia and Hakkani-Tur, Dilek", 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.935", doi = "10.18653/v1/2023.findings-emnlp.935", pages = "14031--14047", abstract = "Dialogue models are able to generate coherent and fluent responses, but they can still be challenging to control and may produce non-engaging, unsafe results. This unpredictability diminishes user trust and can hinder the use of the models in the real world. To address this, we introduce DialGuide, a novel framework for controlling dialogue model behavior using natural language rules, or guidelines. These guidelines provide information about the context they are applicable to and what should be included in the response, allowing the models to generate responses that are more closely aligned with the developer{'}s expectations and intent. We evaluate DialGuide on three tasks in open-domain dialogue response generation: guideline selection, response generation, and response entailment verification. Our dataset contains 10,737 positive and 15,467 negative dialogue context-response-guideline triplets across two domains - chit-chat and safety. We provide baseline models for the tasks and benchmark their performance. We also demonstrate that DialGuide is effective in the dialogue safety domain, producing safe and engaging responses that follow developer guidelines.", }

Dialogue models are able to generate coherent and fluent responses, but they can still be challenging to control and may produce non-engaging, unsafe results. This unpredictability diminishes user trust and can hinder the use of the models in the real world. To address this, we introduce DialGuide, a novel framework for controlling dialogue model behavior using natural language rules, or guidelines. These guidelines provide information about the context they are applicable to and what should be included in the response, allowing the models to generate responses that are more closely aligned with the developer{'}s expectations and intent. We evaluate DialGuide on three tasks in open-domain dialogue response generation: guideline selection, response generation, and response entailment verification. Our dataset contains 10,737 positive and 15,467 negative dialogue context-response-guideline triplets across two domains - chit-chat and safety. We provide baseline models for the tasks and benchmark their performance. We also demonstrate that DialGuide is effective in the dialogue safety domain, producing safe and engaging responses that follow developer guidelines.

DialGuide: Aligning Dialogue Model Behavior with Developer Guidelines

findings-emnlp.935

2212.10557

Poster

https://aclanthology.org/2023.findings-emnlp.936.bib

https://aclanthology.org/2023.findings-emnlp.936/

@inproceedings{peng-etal-2023-rwkv, title = "{RWKV}: Reinventing {RNN}s for the Transformer Era", author = "Peng, Bo and Alcaide, Eric and Anthony, Quentin and Albalak, Alon and Arcadinho, Samuel and Biderman, Stella and Cao, Huanqi and Cheng, Xin and Chung, Michael and Derczynski, Leon and Du, Xingjian and Grella, Matteo and Gv, Kranthi and He, Xuzheng and Hou, Haowen and Kazienko, Przemyslaw and Kocon, Jan and Kong, Jiaming and Koptyra, Bart{\l}omiej and Lau, Hayden and Lin, Jiaju and Mantri, Krishna Sri Ipsit and Mom, Ferdinand and Saito, Atsushi and Song, Guangyu and Tang, Xiangru and Wind, Johan and Wo{\'z}niak, Stanis{\l}aw and Zhang, Zhenyuan and Zhou, Qinghua and Zhu, Jian and Zhu, Rui-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.936", doi = "10.18653/v1/2023.findings-emnlp.936", pages = "14048--14077", abstract = "Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length. In contrast, recurrent neural networks (RNNs) exhibit linear scaling in memory and computational requirements but struggle to match the same performance as Transformers due to limitations in parallelization and scalability. We propose a novel model architecture, Receptance Weighted Key Value (RWKV), that combines the efficient parallelizable training of transformers with the efficient inference of RNNs. Our approach leverages a linear attention mechanism and allows us to formulate the model as either a Transformer or an RNN, thus parallelizing computations during training and maintains constant computational and memory complexity during inference. We scale our models as large as 14 billion parameters, by far the largest dense RNN ever trained, and find RWKV performs on par with similarly sized Transformers, suggesting future work can leverage this architecture to create more efficient models. This work presents a significant step towards reconciling trade-offs between computational efficiency and model performance in sequence processing tasks.", }

Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length. In contrast, recurrent neural networks (RNNs) exhibit linear scaling in memory and computational requirements but struggle to match the same performance as Transformers due to limitations in parallelization and scalability. We propose a novel model architecture, Receptance Weighted Key Value (RWKV), that combines the efficient parallelizable training of transformers with the efficient inference of RNNs. Our approach leverages a linear attention mechanism and allows us to formulate the model as either a Transformer or an RNN, thus parallelizing computations during training and maintains constant computational and memory complexity during inference. We scale our models as large as 14 billion parameters, by far the largest dense RNN ever trained, and find RWKV performs on par with similarly sized Transformers, suggesting future work can leverage this architecture to create more efficient models. This work presents a significant step towards reconciling trade-offs between computational efficiency and model performance in sequence processing tasks.

RWKV: Reinventing RNNs for the Transformer Era

findings-emnlp.936

2305.13048

https://huggingface.co/papers/2305.13048

Poster

https://aclanthology.org/2023.findings-emnlp.937.bib

https://aclanthology.org/2023.findings-emnlp.937/

@inproceedings{hung-etal-2023-wrote, title = "Who Wrote it and Why? Prompting Large-Language Models for Authorship Verification", author = "Hung, Chia-Yu and Hu, Zhiqiang and Hu, Yujia and Lee, Roy", 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.937", doi = "10.18653/v1/2023.findings-emnlp.937", pages = "14078--14084", abstract = "Authorship verification (AV) is a fundamental task in natural language processing (NLP) and computational linguistics, with applications in forensic analysis, plagiarism detection, and identification of deceptive content. Existing AV techniques, including traditional stylometric and deep learning approaches, face limitations in terms of data requirements and lack of explainability. To address these limitations, this paper proposes PromptAV, a novel technique that leverages Large-Language Models (LLMs) for AV by providing step-by-step stylometric explanation prompts. PromptAV outperforms state-of-the-art baselines, operates effectively with limited training data, and enhances interpretability through intuitive explanations, showcasing its potential as an effective and interpretable solution for the AV task.", }

Authorship verification (AV) is a fundamental task in natural language processing (NLP) and computational linguistics, with applications in forensic analysis, plagiarism detection, and identification of deceptive content. Existing AV techniques, including traditional stylometric and deep learning approaches, face limitations in terms of data requirements and lack of explainability. To address these limitations, this paper proposes PromptAV, a novel technique that leverages Large-Language Models (LLMs) for AV by providing step-by-step stylometric explanation prompts. PromptAV outperforms state-of-the-art baselines, operates effectively with limited training data, and enhances interpretability through intuitive explanations, showcasing its potential as an effective and interpretable solution for the AV task.

Who Wrote it and Why? Prompting Large-Language Models for Authorship Verification

findings-emnlp.937

2310.08123

Poster

https://aclanthology.org/2023.findings-emnlp.938.bib

https://aclanthology.org/2023.findings-emnlp.938/

@inproceedings{nguyen-etal-2023-transitioning, title = "Transitioning Representations between Languages for Cross-lingual Event Detection via Langevin Dynamics", author = "Nguyen, Chien and Nguyen, Huy and Dernoncourt, Franck and Nguyen, Thien", 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.938", doi = "10.18653/v1/2023.findings-emnlp.938", pages = "14085--14093", abstract = "Cross-lingual transfer learning (CLTL) for event detection (ED) aims to develop models in high-resource source languages that can be directly applied to produce effective performance for lower-resource target languages. Previous research in this area has focused on representation matching methods to develop a language-universal representation space into which source- and target-language example representations can be mapped to achieve cross-lingual transfer. However, as this approach modifies the representations for the source-language examples, the models might lose discriminative features for ED that are learned over training data of the source language to prevent effective predictions. To this end, our work introduces a novel approach for cross-lingual ED where we only aim to transition the representations for the target-language examples into the source-language space, thus preserving the representations in the source language and their discriminative information. Our method introduces Langevin Dynamics to perform representation transition and a semantic preservation framework to retain event type features during the transition process. Extensive experiments over three languages demonstrate the state-of-the-art performance for ED in CLTL.", }

Cross-lingual transfer learning (CLTL) for event detection (ED) aims to develop models in high-resource source languages that can be directly applied to produce effective performance for lower-resource target languages. Previous research in this area has focused on representation matching methods to develop a language-universal representation space into which source- and target-language example representations can be mapped to achieve cross-lingual transfer. However, as this approach modifies the representations for the source-language examples, the models might lose discriminative features for ED that are learned over training data of the source language to prevent effective predictions. To this end, our work introduces a novel approach for cross-lingual ED where we only aim to transition the representations for the target-language examples into the source-language space, thus preserving the representations in the source language and their discriminative information. Our method introduces Langevin Dynamics to perform representation transition and a semantic preservation framework to retain event type features during the transition process. Extensive experiments over three languages demonstrate the state-of-the-art performance for ED in CLTL.

Transitioning Representations between Languages for Cross-lingual Event Detection via Langevin Dynamics

findings-emnlp.938

Poster

https://aclanthology.org/2023.findings-emnlp.939.bib

https://aclanthology.org/2023.findings-emnlp.939/

@inproceedings{katz-belinkov-2023-visit, title = "{VISIT}: Visualizing and Interpreting the Semantic Information Flow of Transformers", author = "Katz, Shahar and Belinkov, Yonatan", 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.939", doi = "10.18653/v1/2023.findings-emnlp.939", pages = "14094--14113", abstract = "Recent advances in interpretability suggest we can project weights and hidden states of transformer-based language models (LMs) to their vocabulary, a transformation that makes them more human interpretable. In this paper, we investigate LM attention heads and memory values, the vectors the models dynamically create and recall while processing a given input. By analyzing the tokens they represent through this projection, we identify patterns in the information flow inside the attention mechanism. Based on our discoveries, we create a tool to visualize a forward pass of Generative Pre-trained Transformers (GPTs) as an interactive flow graph, with nodes representing neurons or hidden states and edges representing the interactions between them. Our visualization simplifies huge amounts of data into easy-to-read plots that can reflect the models{'} internal processing, uncovering the contribution of each component to the models{'} final prediction. Our visualization also unveils new insights about the role of layer norms as semantic filters that influence the models{'} output, and about neurons that are always activated during forward passes and act as regularization vectors.", }

Recent advances in interpretability suggest we can project weights and hidden states of transformer-based language models (LMs) to their vocabulary, a transformation that makes them more human interpretable. In this paper, we investigate LM attention heads and memory values, the vectors the models dynamically create and recall while processing a given input. By analyzing the tokens they represent through this projection, we identify patterns in the information flow inside the attention mechanism. Based on our discoveries, we create a tool to visualize a forward pass of Generative Pre-trained Transformers (GPTs) as an interactive flow graph, with nodes representing neurons or hidden states and edges representing the interactions between them. Our visualization simplifies huge amounts of data into easy-to-read plots that can reflect the models{'} internal processing, uncovering the contribution of each component to the models{'} final prediction. Our visualization also unveils new insights about the role of layer norms as semantic filters that influence the models{'} output, and about neurons that are always activated during forward passes and act as regularization vectors.

VISIT: Visualizing and Interpreting the Semantic Information Flow of Transformers

findings-emnlp.939

2305.13417

https://huggingface.co/papers/2305.13417

Poster

https://aclanthology.org/2023.findings-emnlp.940.bib

https://aclanthology.org/2023.findings-emnlp.940/

@inproceedings{pan-etal-2023-robustness, title = "Is Robustness Transferable across Languages in Multilingual Neural Machine Translation?", author = "Pan, Leiyu and {Supryadi} and Xiong, Deyi", 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.940", doi = "10.18653/v1/2023.findings-emnlp.940", pages = "14114--14125", abstract = "Robustness, the ability of models to maintain performance in the face of perturbations, is critical for developing reliable NLP systems. Recent studies have shown promising results in improving the robustness of models through adversarial training and data augmentation. However, in machine translation, most of these studies have focused on bilingual machine translation with a single translation direction. In this paper, we investigate the transferability of robustness across different languages in multilingual neural machine translation. We propose a robustness transfer analysis protocol and conduct a series of experiments. In particular, we use character-, word-, and multi-level noises to attack the specific translation direction of the multilingual neural machine translation model and evaluate the robustness of other translation directions. Our findings demonstrate that the robustness gained in one translation direction can indeed transfer to other translation directions. Additionally, we empirically find scenarios where robustness to character-level noise and word-level noise is more likely to transfer.", }

Robustness, the ability of models to maintain performance in the face of perturbations, is critical for developing reliable NLP systems. Recent studies have shown promising results in improving the robustness of models through adversarial training and data augmentation. However, in machine translation, most of these studies have focused on bilingual machine translation with a single translation direction. In this paper, we investigate the transferability of robustness across different languages in multilingual neural machine translation. We propose a robustness transfer analysis protocol and conduct a series of experiments. In particular, we use character-, word-, and multi-level noises to attack the specific translation direction of the multilingual neural machine translation model and evaluate the robustness of other translation directions. Our findings demonstrate that the robustness gained in one translation direction can indeed transfer to other translation directions. Additionally, we empirically find scenarios where robustness to character-level noise and word-level noise is more likely to transfer.

Is Robustness Transferable across Languages in Multilingual Neural Machine Translation?

findings-emnlp.940

2310.20162

Poster

https://aclanthology.org/2023.findings-emnlp.941.bib

https://aclanthology.org/2023.findings-emnlp.941/

@inproceedings{mullappilly-etal-2023-arabic, title = "{A}rabic Mini-{C}limate{GPT} : A Climate Change and Sustainability Tailored {A}rabic {LLM}", author = "Mullappilly, Sahal and Shaker, Abdelrahman and Thawakar, Omkar and Cholakkal, Hisham and Anwer, Rao and Khan, Salman and Khan, Fahad", 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.941", doi = "10.18653/v1/2023.findings-emnlp.941", pages = "14126--14136", abstract = "Climate change is one of the most significant challenges we face together as a society. Creating awareness and educating policy makers the wide-ranging impact of climate change is an essential step towards a sustainable future. Recently, Large Language Models (LLMs) like ChatGPT and Bard have shown impressive conversational abilities and excel in a wide variety of NLP tasks. While these models are close-source, recently alternative open-source LLMs such as Stanford Alpaca and Vicuna have shown promising results. However, these open-source models are not specifically tailored for climate related domain specific information and also struggle to generate meaningful responses in other languages such as, Arabic. To this end, we propose a light-weight Arabic Mini-ClimateGPT that is built on an open-source LLM and is specifically fine-tuned on a conversational-style instruction tuning curated Arabic dataset Clima500-Instruct with over 500k instructions about climate change and sustainability. Further, our model also utilizes a vector embedding based retrieval mechanism during inference. We validate our proposed model through quantitative and qualitative evaluations on climate-related queries. Our model surpasses the baseline LLM in 88.3{\%} of cases during ChatGPT-based evaluation. Furthermore, our human expert evaluation reveals an 81.6{\%} preference for our model{'}s responses over multiple popular open-source models. Our open-source demos, models and curated instruction sets are available here : https://github.com/mbzuai-oryx/ClimateGPT", }

Climate change is one of the most significant challenges we face together as a society. Creating awareness and educating policy makers the wide-ranging impact of climate change is an essential step towards a sustainable future. Recently, Large Language Models (LLMs) like ChatGPT and Bard have shown impressive conversational abilities and excel in a wide variety of NLP tasks. While these models are close-source, recently alternative open-source LLMs such as Stanford Alpaca and Vicuna have shown promising results. However, these open-source models are not specifically tailored for climate related domain specific information and also struggle to generate meaningful responses in other languages such as, Arabic. To this end, we propose a light-weight Arabic Mini-ClimateGPT that is built on an open-source LLM and is specifically fine-tuned on a conversational-style instruction tuning curated Arabic dataset Clima500-Instruct with over 500k instructions about climate change and sustainability. Further, our model also utilizes a vector embedding based retrieval mechanism during inference. We validate our proposed model through quantitative and qualitative evaluations on climate-related queries. Our model surpasses the baseline LLM in 88.3{\%} of cases during ChatGPT-based evaluation. Furthermore, our human expert evaluation reveals an 81.6{\%} preference for our model{'}s responses over multiple popular open-source models. Our open-source demos, models and curated instruction sets are available here : https://github.com/mbzuai-oryx/ClimateGPT

Arabic Mini-ClimateGPT : A Climate Change and Sustainability Tailored Arabic LLM

findings-emnlp.941

2312.09366

Poster

https://aclanthology.org/2023.findings-emnlp.942.bib

https://aclanthology.org/2023.findings-emnlp.942/

@inproceedings{mathur-etal-2023-interpreting, title = "Interpreting Answers to Yes-No Questions in User-Generated Content", author = "Mathur, Shivam and Park, Keun and Chinnappa, Dhivya and Kotamraju, Saketh and Blanco, Eduardo", 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.942", doi = "10.18653/v1/2023.findings-emnlp.942", pages = "14137--14161", abstract = "Interpreting answers to yes-no questions in social media is difficult. Yes and no keywords are uncommon, and the few answers that include them are rarely to be interpreted what the keywords suggest. In this paper, we present a new corpus of 4,442 yes-no question-answer pairs from Twitter. We discuss linguistic characteristics of answers whose interpretation is yes or no, as well as answers whose interpretation is unknown. We show that large language models are far from solving this problem, even after fine-tuning and blending other corpora for the same problem but outside social media.", }

Interpreting answers to yes-no questions in social media is difficult. Yes and no keywords are uncommon, and the few answers that include them are rarely to be interpreted what the keywords suggest. In this paper, we present a new corpus of 4,442 yes-no question-answer pairs from Twitter. We discuss linguistic characteristics of answers whose interpretation is yes or no, as well as answers whose interpretation is unknown. We show that large language models are far from solving this problem, even after fine-tuning and blending other corpora for the same problem but outside social media.

Interpreting Answers to Yes-No Questions in User-Generated Content

findings-emnlp.942

2310.15464

Poster

https://aclanthology.org/2023.findings-emnlp.943.bib

https://aclanthology.org/2023.findings-emnlp.943/

@inproceedings{li-etal-2023-task, title = "Task-Aware Self-Supervised Framework for Dialogue Discourse Parsing", author = "Li, Wei and Zhu, Luyao and Shao, Wei and Yang, Zonglin and Cambria, Erik", 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.943", doi = "10.18653/v1/2023.findings-emnlp.943", pages = "14162--14173", abstract = "Dialogue discourse parsing is a fundamental natural language processing task. It can benefit a series of conversation-related downstream tasks including dialogue summarization and emotion recognition in conversations. However, existing parsing approaches are constrained by predefined relation types, which can impede the adaptability of the parser for downstream tasks. To this end, we propose to introduce a task-aware paradigm to improve the versatility of the parser in this paper. Moreover, to alleviate error propagation and learning bias, we design a graph-based discourse parsing model termed DialogDP. Building upon the symmetrical property of matrix-embedded parsing graphs, we have developed an innovative self-supervised mechanism that leverages both bottom-up and top-down parsing strategies. This approach allows the parsing graphs to mutually regularize and enhance each other. Empirical studies on dialogue discourse parsing datasets and a downstream task demonstrate the effectiveness and flexibility of our framework.", }

Dialogue discourse parsing is a fundamental natural language processing task. It can benefit a series of conversation-related downstream tasks including dialogue summarization and emotion recognition in conversations. However, existing parsing approaches are constrained by predefined relation types, which can impede the adaptability of the parser for downstream tasks. To this end, we propose to introduce a task-aware paradigm to improve the versatility of the parser in this paper. Moreover, to alleviate error propagation and learning bias, we design a graph-based discourse parsing model termed DialogDP. Building upon the symmetrical property of matrix-embedded parsing graphs, we have developed an innovative self-supervised mechanism that leverages both bottom-up and top-down parsing strategies. This approach allows the parsing graphs to mutually regularize and enhance each other. Empirical studies on dialogue discourse parsing datasets and a downstream task demonstrate the effectiveness and flexibility of our framework.

Task-Aware Self-Supervised Framework for Dialogue Discourse Parsing

findings-emnlp.943

Poster

https://aclanthology.org/2023.findings-emnlp.944.bib

https://aclanthology.org/2023.findings-emnlp.944/

@inproceedings{chang-fosler-lussier-2023-selective, title = "Selective Demonstrations for Cross-domain Text-to-{SQL}", author = "Chang, Shuaichen and Fosler-Lussier, Eric", 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.944", doi = "10.18653/v1/2023.findings-emnlp.944", pages = "14174--14189", abstract = "Large language models (LLMs) with in-context learning have demonstrated impressive generalization capabilities in the cross-domain text-to-SQL task, without the use of in-domain annotations. However, incorporating in-domain demonstration examples has been found to greatly enhance LLMs{'} performance. In this paper, we delve into the key factors within in-domain examples that contribute to the improvement and explore whether we can harness these benefits without relying on in-domain annotations. Based on our findings, we propose a demonstration selection framework, ODIS, which utilizes both out-of-domain examples and synthetically generated in-domain examples to construct demonstrations. By retrieving demonstrations from hybrid sources, ODIS leverages the advantages of both, showcasing its effectiveness compared to baseline methods that rely on a single data source. Furthermore, ODIS outperforms state-of-the-art approaches on two cross-domain text-to-SQL datasets, with improvements of 1.1 and 11.8 points in execution accuracy, respectively.", }

Large language models (LLMs) with in-context learning have demonstrated impressive generalization capabilities in the cross-domain text-to-SQL task, without the use of in-domain annotations. However, incorporating in-domain demonstration examples has been found to greatly enhance LLMs{'} performance. In this paper, we delve into the key factors within in-domain examples that contribute to the improvement and explore whether we can harness these benefits without relying on in-domain annotations. Based on our findings, we propose a demonstration selection framework, ODIS, which utilizes both out-of-domain examples and synthetically generated in-domain examples to construct demonstrations. By retrieving demonstrations from hybrid sources, ODIS leverages the advantages of both, showcasing its effectiveness compared to baseline methods that rely on a single data source. Furthermore, ODIS outperforms state-of-the-art approaches on two cross-domain text-to-SQL datasets, with improvements of 1.1 and 11.8 points in execution accuracy, respectively.

Selective Demonstrations for Cross-domain Text-to-SQL

findings-emnlp.944

2310.06302

Poster

https://aclanthology.org/2023.findings-emnlp.945.bib

https://aclanthology.org/2023.findings-emnlp.945/

@inproceedings{wang-izbicki-2023-docsplit, title = "{D}oc{S}plit: Simple Contrastive Pretraining for Large Document Embeddings", author = "Wang, Yujie and Izbicki, Mike", 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.945", doi = "10.18653/v1/2023.findings-emnlp.945", pages = "14190--14196", abstract = "Existing model pretraining methods only consider local information. For example, in the popular token masking strategy, the words closer to the masked token are more important for prediction than words far away. This results in pretrained models that generate high-quality sentence embeddings, but low-quality embeddings for large documents. We propose a new pretraining method called DocSplit which forces models to consider the entire global context of a large document. Our method uses a contrastive loss where the positive examples are randomly sampled sections of the input document, and negative examples are randomly sampled sections of unrelated documents. Like previous pretraining methods, DocSplit is fully unsupervised, easy to implement, and can be used to pretrain any model architecture. Our experiments show that DocSplit outperforms other pretraining methods for document classification, few shot learning, and information retrieval tasks.", }

Existing model pretraining methods only consider local information. For example, in the popular token masking strategy, the words closer to the masked token are more important for prediction than words far away. This results in pretrained models that generate high-quality sentence embeddings, but low-quality embeddings for large documents. We propose a new pretraining method called DocSplit which forces models to consider the entire global context of a large document. Our method uses a contrastive loss where the positive examples are randomly sampled sections of the input document, and negative examples are randomly sampled sections of unrelated documents. Like previous pretraining methods, DocSplit is fully unsupervised, easy to implement, and can be used to pretrain any model architecture. Our experiments show that DocSplit outperforms other pretraining methods for document classification, few shot learning, and information retrieval tasks.

DocSplit: Simple Contrastive Pretraining for Large Document Embeddings

findings-emnlp.945

Poster

https://aclanthology.org/2023.findings-emnlp.946.bib

https://aclanthology.org/2023.findings-emnlp.946/

@inproceedings{karmaker-santu-feng-2023-teler, title = "{TEL}e{R}: A General Taxonomy of {LLM} Prompts for Benchmarking Complex Tasks", author = "Karmaker Santu, Shubhra Kanti and Feng, Dongji", 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.946", doi = "10.18653/v1/2023.findings-emnlp.946", pages = "14197--14203", abstract = "While LLMs have shown great success in understanding and generating text in traditional conversational settings, their potential for performing ill-defined complex tasks is largely under-studied and yet to be benchmarked. However, conducting such benchmarking studies is challenging because of the large variations in LLMs{'} performance when different prompt types/styles are used and different degrees of detail are provided in the prompts. To address this issue, this paper proposes a general taxonomy that can be used to design prompts with specific properties in order to perform a wide range of complex tasks. This taxonomy will allow future benchmarking studies to report the specific categories of prompts used as part of the study, enabling meaningful comparisons across different studies. Also, by establishing a common standard through this taxonomy, researchers will be able to draw more accurate conclusions about LLMs{'} performance on a specific complex task.", }

While LLMs have shown great success in understanding and generating text in traditional conversational settings, their potential for performing ill-defined complex tasks is largely under-studied and yet to be benchmarked. However, conducting such benchmarking studies is challenging because of the large variations in LLMs{'} performance when different prompt types/styles are used and different degrees of detail are provided in the prompts. To address this issue, this paper proposes a general taxonomy that can be used to design prompts with specific properties in order to perform a wide range of complex tasks. This taxonomy will allow future benchmarking studies to report the specific categories of prompts used as part of the study, enabling meaningful comparisons across different studies. Also, by establishing a common standard through this taxonomy, researchers will be able to draw more accurate conclusions about LLMs{'} performance on a specific complex task.

TELeR: A General Taxonomy of LLM Prompts for Benchmarking Complex Tasks

findings-emnlp.946

2305.11430

Poster

https://aclanthology.org/2023.findings-emnlp.947.bib

https://aclanthology.org/2023.findings-emnlp.947/

@inproceedings{singhal-etal-2023-intendd, title = "{I}nten{DD}: A Unified Contrastive Learning Approach for Intent Detection and Discovery", author = "Singhal, Bhavuk and Gupta, Ashim and Shivasankaran, V P and Krishna, Amrith", 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.947", doi = "10.18653/v1/2023.findings-emnlp.947", pages = "14204--14216", abstract = "Identifying intents from dialogue utterances forms an integral component of task-oriented dialogue systems. Intent-related tasks are typically formulated either as a classification task, where the utterances are classified into predefined categories or as a clustering task when new and previously unknown intent categories need to be discovered from these utterances. Further, the intent classification may be modeled in a multiclass (MC) or multilabel (ML) setup. While typically these tasks are modeled as separate tasks, we propose IntenDD a unified approach leveraging a shared utterance encoding backbone. IntenDD uses an entirely unsupervised contrastive learning strategy for representation learning, where pseudo-labels for the unlabeled utterances are generated based on their lexical features. Additionally, we introduce a two-step post-processing setup for the classification tasks using modified adsorption. Here, first, the residuals in the training data are propagated followed by smoothing the labels both modeled in a transductive setting. Through extensive evaluations on various benchmark datasets, we find that our approach consistently outperforms competitive baselines across all three tasks. On average, IntenDD reports percentage improvements of 2.32 {\%}, 1.26 {\%}, and 1.52 {\%} in their respective metrics for few-shot MC, few-shot ML, and the intent discovery tasks respectively.", }

Identifying intents from dialogue utterances forms an integral component of task-oriented dialogue systems. Intent-related tasks are typically formulated either as a classification task, where the utterances are classified into predefined categories or as a clustering task when new and previously unknown intent categories need to be discovered from these utterances. Further, the intent classification may be modeled in a multiclass (MC) or multilabel (ML) setup. While typically these tasks are modeled as separate tasks, we propose IntenDD a unified approach leveraging a shared utterance encoding backbone. IntenDD uses an entirely unsupervised contrastive learning strategy for representation learning, where pseudo-labels for the unlabeled utterances are generated based on their lexical features. Additionally, we introduce a two-step post-processing setup for the classification tasks using modified adsorption. Here, first, the residuals in the training data are propagated followed by smoothing the labels both modeled in a transductive setting. Through extensive evaluations on various benchmark datasets, we find that our approach consistently outperforms competitive baselines across all three tasks. On average, IntenDD reports percentage improvements of 2.32 {\%}, 1.26 {\%}, and 1.52 {\%} in their respective metrics for few-shot MC, few-shot ML, and the intent discovery tasks respectively.

IntenDD: A Unified Contrastive Learning Approach for Intent Detection and Discovery

findings-emnlp.947

2310.16761

Poster

https://aclanthology.org/2023.findings-emnlp.948.bib

https://aclanthology.org/2023.findings-emnlp.948/

@inproceedings{shang-etal-2023-inarig, title = "{IN}ar{IG}: Iterative Non-autoregressive Instruct Generation Model For Word-Level Auto Completion", author = "Shang, Hengchao and Li, Zongyao and Wei, Daimeng and Guo, Jiaxin and Wang, Minghan and Chen, Xiaoyu and Lei, Lizhi and Yang, Hao", 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.948", doi = "10.18653/v1/2023.findings-emnlp.948", pages = "14217--14228", abstract = "Computer-aided translation (CAT) aims to enhance human translation efficiency and is still important in scenarios where machine translation cannot meet quality requirements. One fundamental task within this field is Word-Level Auto Completion (WLAC). WLAC predicts a target word given a source sentence, translation context, and a human typed character sequence. Previous works either employ word classification models to exploit contextual information from both sides of the target word or directly disregarded the dependencies from the right-side context. Furthermore, the key information, i.e. human typed sequences, is only used as prefix constraints in the decoding module. In this paper, we propose the INarIG (Iterative Non-autoregressive Instruct Generation) model, which constructs the human typed sequence into Instruction Unit and employs iterative decoding with subwords to fully utilize input information given in the task. Our model is more competent in dealing with low-frequency words (core scenario of this task), and achieves state-of-the-art results on the WMT22 and benchmark datasets, with a maximum increase of over 10{\%} prediction accuracy.", }

Computer-aided translation (CAT) aims to enhance human translation efficiency and is still important in scenarios where machine translation cannot meet quality requirements. One fundamental task within this field is Word-Level Auto Completion (WLAC). WLAC predicts a target word given a source sentence, translation context, and a human typed character sequence. Previous works either employ word classification models to exploit contextual information from both sides of the target word or directly disregarded the dependencies from the right-side context. Furthermore, the key information, i.e. human typed sequences, is only used as prefix constraints in the decoding module. In this paper, we propose the INarIG (Iterative Non-autoregressive Instruct Generation) model, which constructs the human typed sequence into Instruction Unit and employs iterative decoding with subwords to fully utilize input information given in the task. Our model is more competent in dealing with low-frequency words (core scenario of this task), and achieves state-of-the-art results on the WMT22 and benchmark datasets, with a maximum increase of over 10{\%} prediction accuracy.

INarIG: Iterative Non-autoregressive Instruct Generation Model For Word-Level Auto Completion

findings-emnlp.948

2311.18200

Poster

https://aclanthology.org/2023.findings-emnlp.949.bib

https://aclanthology.org/2023.findings-emnlp.949/

@inproceedings{henning-etal-2023-answer, title = "Is the Answer in the Text? Challenging {C}hat{GPT} with Evidence Retrieval from Instructive Text", author = "Henning, Sophie and Anthonio, Talita and Zhou, Wei and Adel, Heike and Mesgar, Mohsen and Friedrich, Annemarie", 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.949", pages = "14229--14241", abstract = "Generative language models have recently shown remarkable success in generating answers to questions in a given textual context. However, these answers may suffer from hallucination, wrongly cite evidence, and spread misleading information. In this work, we address this problem by employing ChatGPT, a state-of-the-art generative model, as a machine-reading system. We ask it to retrieve answers to lexically varied and open-ended questions from trustworthy instructive texts. We introduce WHERE (WikiHow Evidence REtrieval), a new high-quality evaluation benchmark of a set of WikiHow articles exhaustively annotated with evidence sentences to questions that comes with a special challenge: All questions are about the article{'}s topic, but not all can be answered using the provided context. We interestingly find that when using a regular question-answering prompt, ChatGPT neglects to detect the unanswerable cases. When provided with a few examples, it learns to better judge whether a text provides answer evidence or not. Alongside this important finding, our dataset defines a new benchmark for evidence retrieval in question answering, which we argue is one of the necessary next steps for making large language models more trustworthy.", }

Generative language models have recently shown remarkable success in generating answers to questions in a given textual context. However, these answers may suffer from hallucination, wrongly cite evidence, and spread misleading information. In this work, we address this problem by employing ChatGPT, a state-of-the-art generative model, as a machine-reading system. We ask it to retrieve answers to lexically varied and open-ended questions from trustworthy instructive texts. We introduce WHERE (WikiHow Evidence REtrieval), a new high-quality evaluation benchmark of a set of WikiHow articles exhaustively annotated with evidence sentences to questions that comes with a special challenge: All questions are about the article{'}s topic, but not all can be answered using the provided context. We interestingly find that when using a regular question-answering prompt, ChatGPT neglects to detect the unanswerable cases. When provided with a few examples, it learns to better judge whether a text provides answer evidence or not. Alongside this important finding, our dataset defines a new benchmark for evidence retrieval in question answering, which we argue is one of the necessary next steps for making large language models more trustworthy.

Is the Answer in the Text? Challenging ChatGPT with Evidence Retrieval from Instructive Text

findings-emnlp.949

Poster

https://aclanthology.org/2023.findings-emnlp.950.bib

https://aclanthology.org/2023.findings-emnlp.950/

@inproceedings{drozdov-etal-2023-parade, title = "{P}a{R}a{D}e: Passage Ranking using Demonstrations with {LLM}s", author = "Drozdov, Andrew and Zhuang, Honglei and Dai, Zhuyun and Qin, Zhen and Rahimi, Razieh and Wang, Xuanhui and Alon, Dana and Iyyer, Mohit and McCallum, Andrew and Metzler, Donald and Hui, Kai", 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.950", doi = "10.18653/v1/2023.findings-emnlp.950", pages = "14242--14252", abstract = "Recent studies show that large language models (LLMs) can be instructed to effectively perform zero-shot passage re-ranking, in which the results of a first stage retrieval method, such as BM25, are rated and reordered to improve relevance. In this work, we improve LLM-based re-ranking by algorithmically selecting few-shot demonstrations to include in the prompt. Our analysis investigates the conditions where demonstrations are most helpful, and shows that adding even one demonstration is significantly beneficial. We propose a novel demonstration selection strategy based on difficulty rather than the commonly used semantic similarity. Furthermore, we find that demonstrations helpful for ranking are also effective at question generation. We hope our work will spur more principled research into question generation and passage ranking.", }

Recent studies show that large language models (LLMs) can be instructed to effectively perform zero-shot passage re-ranking, in which the results of a first stage retrieval method, such as BM25, are rated and reordered to improve relevance. In this work, we improve LLM-based re-ranking by algorithmically selecting few-shot demonstrations to include in the prompt. Our analysis investigates the conditions where demonstrations are most helpful, and shows that adding even one demonstration is significantly beneficial. We propose a novel demonstration selection strategy based on difficulty rather than the commonly used semantic similarity. Furthermore, we find that demonstrations helpful for ranking are also effective at question generation. We hope our work will spur more principled research into question generation and passage ranking.

PaRaDe: Passage Ranking using Demonstrations with LLMs

findings-emnlp.950

Poster

https://aclanthology.org/2023.findings-emnlp.951.bib

https://aclanthology.org/2023.findings-emnlp.951/

@inproceedings{liu-etal-2023-learning-dynamic, title = "Learning Dynamic Representations for Discourse Dependency Parsing", author = "Liu, Tianyi and Feng, Yansong and Zhao, Dongyan", 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.951", doi = "10.18653/v1/2023.findings-emnlp.951", pages = "14253--14263", abstract = "Transition systems have been widely used for the discourse dependency parsing task. Existing works often characterize transition states by examining a certain number of elementary discourse units (EDUs), while neglecting the arcs obtained from the transition history. In this paper, we propose to employ GAT-based encoder to learn dynamic representations for sub-trees constructed in previous transition steps. By incorporating these representations, our model is able to retain accessibility to all parsed EDUs through the obtained arcs, thus better utilizing the structural information of the document, particularly when handling lengthy text spans with complex structures. For the discourse relation recognition task, we employ edge-featured GATs to derive better representations for EDU pairs. Experimental results show that our model can achieve state-of-the-art performance on widely adopted datasets including RST-DT, SciDTB and CDTB. Our code is available at ${https://github.com/lty-lty/Discourse-Dependency-Parsing}$.", }

Transition systems have been widely used for the discourse dependency parsing task. Existing works often characterize transition states by examining a certain number of elementary discourse units (EDUs), while neglecting the arcs obtained from the transition history. In this paper, we propose to employ GAT-based encoder to learn dynamic representations for sub-trees constructed in previous transition steps. By incorporating these representations, our model is able to retain accessibility to all parsed EDUs through the obtained arcs, thus better utilizing the structural information of the document, particularly when handling lengthy text spans with complex structures. For the discourse relation recognition task, we employ edge-featured GATs to derive better representations for EDU pairs. Experimental results show that our model can achieve state-of-the-art performance on widely adopted datasets including RST-DT, SciDTB and CDTB. Our code is available at ${https://github.com/lty-lty/Discourse-Dependency-Parsing}$.

Learning Dynamic Representations for Discourse Dependency Parsing

findings-emnlp.951

Poster

https://aclanthology.org/2023.findings-emnlp.952.bib

https://aclanthology.org/2023.findings-emnlp.952/

@inproceedings{park-etal-2023-k, title = "K-{HATERS}: A Hate Speech Detection Corpus in {K}orean with Target-Specific Ratings", author = "Park, Chaewon and Kim, Soohwan and Park, Kyubyong and Park, Kunwoo", 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.952", doi = "10.18653/v1/2023.findings-emnlp.952", pages = "14264--14278", abstract = "Numerous datasets have been proposed to combat the spread of online hate. Despite these efforts, a majority of these resources are English-centric, primarily focusing on overt forms of hate. This research gap calls for developing high-quality corpora in diverse languages that also encapsulate more subtle hate expressions. This study introduces K-HATERS, a new corpus for hate speech detection in Korean, comprising approximately 192K news comments with target-specific offensiveness ratings. This resource is the largest offensive language corpus in Korean and is the first to offer target-specific ratings on a three-point Likert scale, enabling the detection of hate expressions in Korean across varying degrees of offensiveness. We conduct experiments showing the effectiveness of the proposed corpus, including a comparison with existing datasets. Additionally, to address potential noise and bias in human annotations, we explore a novel idea of adopting the Cognitive Reflection Test, which is widely used in social science for assessing an individual{'}s cognitive ability, as a proxy of labeling quality. Findings indicate that annotations from individuals with the lowest test scores tend to yield detection models that make biased predictions toward specific target groups and are less accurate. This study contributes to the NLP research on hate speech detection and resource construction. The code and dataset can be accessed at https://github.com/ssu-humane/K-HATERS.", }

Numerous datasets have been proposed to combat the spread of online hate. Despite these efforts, a majority of these resources are English-centric, primarily focusing on overt forms of hate. This research gap calls for developing high-quality corpora in diverse languages that also encapsulate more subtle hate expressions. This study introduces K-HATERS, a new corpus for hate speech detection in Korean, comprising approximately 192K news comments with target-specific offensiveness ratings. This resource is the largest offensive language corpus in Korean and is the first to offer target-specific ratings on a three-point Likert scale, enabling the detection of hate expressions in Korean across varying degrees of offensiveness. We conduct experiments showing the effectiveness of the proposed corpus, including a comparison with existing datasets. Additionally, to address potential noise and bias in human annotations, we explore a novel idea of adopting the Cognitive Reflection Test, which is widely used in social science for assessing an individual{'}s cognitive ability, as a proxy of labeling quality. Findings indicate that annotations from individuals with the lowest test scores tend to yield detection models that make biased predictions toward specific target groups and are less accurate. This study contributes to the NLP research on hate speech detection and resource construction. The code and dataset can be accessed at https://github.com/ssu-humane/K-HATERS.

K-HATERS: A Hate Speech Detection Corpus in Korean with Target-Specific Ratings

findings-emnlp.952

2310.15439

https://huggingface.co/papers/2310.15439

Poster

https://aclanthology.org/2023.findings-emnlp.953.bib

https://aclanthology.org/2023.findings-emnlp.953/

@inproceedings{lai-etal-2023-mitigating, title = "Mitigating Data Imbalance and Representation Degeneration in Multilingual Machine Translation", author = "Lai, Wen and Chronopoulou, Alexandra and Fraser, Alexander", 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.953", doi = "10.18653/v1/2023.findings-emnlp.953", pages = "14279--14294", abstract = "Despite advances in multilingual neural machine translation (MNMT), we argue that there are still two major challenges in this area: data imbalance and representation degeneration. The data imbalance problem refers to the imbalance in the amount of parallel corpora for all language pairs, especially for long-tail languages (i.e., very low-resource languages). The representation degeneration problem refers to the problem of encoded tokens tending to appear only in a small subspace of the full space available to the MNMT model. To solve these two issues, we propose Bi-ACL, a framework which only requires target-side monolingual data and a bilingual dictionary to improve the performance of the MNMT model. We define two modules, named bidirectional autoencoder and bidirectional contrastive learning, which we combine with an online constrained beam search and a curriculum learning sampling strategy. Extensive experiments show that our proposed method is more effective than strong baselines both in long-tail languages and in high-resource languages. We also demonstrate that our approach is capable of transferring knowledge between domains and languages in zero-shot scenarios.", }

Despite advances in multilingual neural machine translation (MNMT), we argue that there are still two major challenges in this area: data imbalance and representation degeneration. The data imbalance problem refers to the imbalance in the amount of parallel corpora for all language pairs, especially for long-tail languages (i.e., very low-resource languages). The representation degeneration problem refers to the problem of encoded tokens tending to appear only in a small subspace of the full space available to the MNMT model. To solve these two issues, we propose Bi-ACL, a framework which only requires target-side monolingual data and a bilingual dictionary to improve the performance of the MNMT model. We define two modules, named bidirectional autoencoder and bidirectional contrastive learning, which we combine with an online constrained beam search and a curriculum learning sampling strategy. Extensive experiments show that our proposed method is more effective than strong baselines both in long-tail languages and in high-resource languages. We also demonstrate that our approach is capable of transferring knowledge between domains and languages in zero-shot scenarios.

Mitigating Data Imbalance and Representation Degeneration in Multilingual Machine Translation

findings-emnlp.953

2305.12786

Poster

https://aclanthology.org/2023.findings-emnlp.954.bib

https://aclanthology.org/2023.findings-emnlp.954/

@inproceedings{tan-etal-2023-botpercent, title = "{B}ot{P}ercent: Estimating Bot Populations in {T}witter Communities", author = "Tan, Zhaoxuan and Feng, Shangbin and Sclar, Melanie and Wan, Herun and Luo, Minnan and Choi, Yejin and Tsvetkov, Yulia", 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.954", doi = "10.18653/v1/2023.findings-emnlp.954", pages = "14295--14312", abstract = "Twitter bot detection is vital in combating misinformation and safeguarding the integrity of social media discourse. While malicious bots are becoming more and more sophisticated and personalized, standard bot detection approaches are still agnostic to social environments (henceforth, communities) the bots operate at. In this work, we introduce community-specific bot detection, estimating the percentage of bots given the context of a community. Our method{---}BotPercent{---}is an amalgamation of Twitter bot detection datasets and feature-, text-, and graph-based models, adjusted to a particular community on Twitter. We introduce an approach that performs confidence calibration across bot detection models, which addresses generalization issues in existing community-agnostic models targeting individual bots and leads to more accurate community-level bot estimations. Experiments demonstrate that BotPercent achieves state-of-the-art performance in community-level Twitter bot detection across both balanced and imbalanced class distribution settings, presenting a less biased estimator of Twitter bot populations within the communities we analyze. We then analyze bot rates in several Twitter groups, including users who engage with partisan news media, political communities in different countries, and more. Our results reveal that the presence of Twitter bots is not homogeneous, but exhibiting a spatial-temporal distribution with considerable heterogeneity that should be taken into account for content moderation and social media policy making. The implementation of BotPercent is available at https://github.com/TamSiuhin/BotPercent.", }

Twitter bot detection is vital in combating misinformation and safeguarding the integrity of social media discourse. While malicious bots are becoming more and more sophisticated and personalized, standard bot detection approaches are still agnostic to social environments (henceforth, communities) the bots operate at. In this work, we introduce community-specific bot detection, estimating the percentage of bots given the context of a community. Our method{---}BotPercent{---}is an amalgamation of Twitter bot detection datasets and feature-, text-, and graph-based models, adjusted to a particular community on Twitter. We introduce an approach that performs confidence calibration across bot detection models, which addresses generalization issues in existing community-agnostic models targeting individual bots and leads to more accurate community-level bot estimations. Experiments demonstrate that BotPercent achieves state-of-the-art performance in community-level Twitter bot detection across both balanced and imbalanced class distribution settings, presenting a less biased estimator of Twitter bot populations within the communities we analyze. We then analyze bot rates in several Twitter groups, including users who engage with partisan news media, political communities in different countries, and more. Our results reveal that the presence of Twitter bots is not homogeneous, but exhibiting a spatial-temporal distribution with considerable heterogeneity that should be taken into account for content moderation and social media policy making. The implementation of BotPercent is available at https://github.com/TamSiuhin/BotPercent.

BotPercent: Estimating Bot Populations in Twitter Communities

findings-emnlp.954

2302.00381

Poster

https://aclanthology.org/2023.findings-emnlp.955.bib

https://aclanthology.org/2023.findings-emnlp.955/

@inproceedings{chen-etal-2023-locality, title = "The Locality and Symmetry of Positional Encodings", author = "Chen, Lihu and Varoquaux, Gael and Suchanek, Fabian", 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.955", doi = "10.18653/v1/2023.findings-emnlp.955", pages = "14313--14331", abstract = "Positional Encodings (PEs) are used to inject word-order information into transformer-based language models. While they can significantly enhance the quality of sentence representations, their specific contribution to language models is not fully understood, especially given recent findings that various positional encodings are insensitive to word order. In this work, we conduct a systematic study of positional encodings in \textbf{Bidirectional Masked Language Models} (BERT-style) , which complements existing work in three aspects: (1) We uncover the core function of PEs by identifying two common properties, Locality and Symmetry; (2) We show that the two properties are closely correlated with the performances of downstream tasks; (3) We quantify the weakness of current PEs by introducing two new probing tasks, on which current PEs perform poorly. We believe that these results are the basis for developing better PEs for transformer-based language models.", }

Positional Encodings (PEs) are used to inject word-order information into transformer-based language models. While they can significantly enhance the quality of sentence representations, their specific contribution to language models is not fully understood, especially given recent findings that various positional encodings are insensitive to word order. In this work, we conduct a systematic study of positional encodings in \textbf{Bidirectional Masked Language Models} (BERT-style) , which complements existing work in three aspects: (1) We uncover the core function of PEs by identifying two common properties, Locality and Symmetry; (2) We show that the two properties are closely correlated with the performances of downstream tasks; (3) We quantify the weakness of current PEs by introducing two new probing tasks, on which current PEs perform poorly. We believe that these results are the basis for developing better PEs for transformer-based language models.

The Locality and Symmetry of Positional Encodings

findings-emnlp.955

2310.12864

Poster

https://aclanthology.org/2023.findings-emnlp.956.bib

https://aclanthology.org/2023.findings-emnlp.956/

@inproceedings{sun-etal-2023-towards-deep, title = "Towards a Deep Understanding of Multilingual End-to-End Speech Translation", author = "Sun, Haoran and Zhao, Xiaohu and Lei, Yikun and Zhu, Shaolin and Xiong, Deyi", 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.956", doi = "10.18653/v1/2023.findings-emnlp.956", pages = "14332--14348", abstract = "In this paper, we employ Singular Value Canonical Correlation Analysis (SVCCA) to analyze representations learnt in a multilingual end-to-end speech translation model trained over 22 languages. SVCCA enables us to estimate representational similarity across languages and layers, enhancing our understanding of the functionality of multilingual speech translation and its potential connection to multilingual neural machine translation. The multilingual speech translation model is trained on the CoVoST 2 dataset in all possible directions, and we utilize LASER to extract parallel bitext data for SVCCA analysis. We derive three major findings from our analysis: (I) Linguistic similarity loses its efficacy in multilingual speech translation when the training data for a specific language is limited. (II) Enhanced encoder representations and well-aligned audio-text data significantly improve translation quality, surpassing the bilingual counterparts when the training data is not compromised. (III) The encoder representations of multilingual speech translation demonstrate superior performance in predicting phonetic features in linguistic typology prediction. With these findings, we propose that releasing the constraint of limited data for low-resource languages and subsequently combining them with linguistically related high-resource languages could offer a more effective approach for multilingual end-to-end speech translation.", }

In this paper, we employ Singular Value Canonical Correlation Analysis (SVCCA) to analyze representations learnt in a multilingual end-to-end speech translation model trained over 22 languages. SVCCA enables us to estimate representational similarity across languages and layers, enhancing our understanding of the functionality of multilingual speech translation and its potential connection to multilingual neural machine translation. The multilingual speech translation model is trained on the CoVoST 2 dataset in all possible directions, and we utilize LASER to extract parallel bitext data for SVCCA analysis. We derive three major findings from our analysis: (I) Linguistic similarity loses its efficacy in multilingual speech translation when the training data for a specific language is limited. (II) Enhanced encoder representations and well-aligned audio-text data significantly improve translation quality, surpassing the bilingual counterparts when the training data is not compromised. (III) The encoder representations of multilingual speech translation demonstrate superior performance in predicting phonetic features in linguistic typology prediction. With these findings, we propose that releasing the constraint of limited data for low-resource languages and subsequently combining them with linguistically related high-resource languages could offer a more effective approach for multilingual end-to-end speech translation.

Towards a Deep Understanding of Multilingual End-to-End Speech Translation

findings-emnlp.956

2310.20456

Poster

https://aclanthology.org/2023.findings-emnlp.957.bib

https://aclanthology.org/2023.findings-emnlp.957/

@inproceedings{luo-etal-2023-empirical, title = "An Empirical Investigation of Implicit and Explicit Knowledge-Enhanced Methods for Ad Hoc Dataset Retrieval", author = "Luo, Weiqing and Chen, Qiaosheng and Zhang, Zhiyang and Huang, Zixian and Cheng, Gong", 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.957", doi = "10.18653/v1/2023.findings-emnlp.957", pages = "14349--14360", abstract = "Ad hoc dataset retrieval has become an important way of finding data on the Web, where the underlying problem is how to measure the relevance of a dataset to a query. State-of-the-art solutions for this task are still lexical methods, which cannot capture semantic similarity. Semantics-aware knowledge-enhanced retrieval methods, which achieved promising results on other tasks, have yet to be systematically studied on this specialized task. To fill the gap, in this paper, we present an empirical investigation of the task where we implement and evaluate, on two test collections, a set of implicit and explicit knowledge-enhancement retrieval methods in various settings. Our results reveal the unique features of the task and suggest an interpolation of different kinds of methods as the current best practice.", }

Ad hoc dataset retrieval has become an important way of finding data on the Web, where the underlying problem is how to measure the relevance of a dataset to a query. State-of-the-art solutions for this task are still lexical methods, which cannot capture semantic similarity. Semantics-aware knowledge-enhanced retrieval methods, which achieved promising results on other tasks, have yet to be systematically studied on this specialized task. To fill the gap, in this paper, we present an empirical investigation of the task where we implement and evaluate, on two test collections, a set of implicit and explicit knowledge-enhancement retrieval methods in various settings. Our results reveal the unique features of the task and suggest an interpolation of different kinds of methods as the current best practice.

An Empirical Investigation of Implicit and Explicit Knowledge-Enhanced Methods for Ad Hoc Dataset Retrieval

findings-emnlp.957

Poster

https://aclanthology.org/2023.findings-emnlp.958.bib

https://aclanthology.org/2023.findings-emnlp.958/

@inproceedings{fujinuma-etal-2023-multi, title = "A Multi-Modal Multilingual Benchmark for Document Image Classification", author = "Fujinuma, Yoshinari and Varia, Siddharth and Sankaran, Nishant and Appalaraju, Srikar and Min, Bonan and Vyas, Yogarshi", 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.958", doi = "10.18653/v1/2023.findings-emnlp.958", pages = "14361--14376", abstract = "Document image classification is different from plain-text document classification and consists of classifying a document by understanding the content and structure of documents such as forms, emails, and other such documents. We show that the only existing dataset for this task (Lewis et al., 2006) has several limitations and we introduce two newly curated multilingual datasets WIKI-DOC and MULTIEURLEX-DOC that overcome these limitations. We further undertake a comprehensive study of popular visually-rich document understanding or Document AI models in previously untested setting in document image classification such as 1) multi-label classification, and 2) zero-shot cross-lingual transfer setup. Experimental results show limitations of multilingual Document AI models on cross-lingual transfer across typologically distant languages. Our datasets and findings open the door for future research into improving Document AI models.", }

Document image classification is different from plain-text document classification and consists of classifying a document by understanding the content and structure of documents such as forms, emails, and other such documents. We show that the only existing dataset for this task (Lewis et al., 2006) has several limitations and we introduce two newly curated multilingual datasets WIKI-DOC and MULTIEURLEX-DOC that overcome these limitations. We further undertake a comprehensive study of popular visually-rich document understanding or Document AI models in previously untested setting in document image classification such as 1) multi-label classification, and 2) zero-shot cross-lingual transfer setup. Experimental results show limitations of multilingual Document AI models on cross-lingual transfer across typologically distant languages. Our datasets and findings open the door for future research into improving Document AI models.

A Multi-Modal Multilingual Benchmark for Document Image Classification

findings-emnlp.958

2310.16356

Poster

https://aclanthology.org/2023.findings-emnlp.959.bib

https://aclanthology.org/2023.findings-emnlp.959/

@inproceedings{kervadec-etal-2023-unnatural, title = "Unnatural language processing: How do language models handle machine-generated prompts?", author = "Kervadec, Corentin and Franzon, Francesca and Baroni, Marco", 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.959", doi = "10.18653/v1/2023.findings-emnlp.959", pages = "14377--14392", abstract = "Language model prompt optimization research has shown that semantically and grammatically well-formed manually crafted prompts are routinely outperformed by automatically generated token sequences with no apparent meaning or syntactic structure, including sequences of vectors from a model{'}s embedding space. We use machine-generated prompts to probe how models respond to input that is not composed of natural language expressions. We study the behavior of models of different sizes in multiple semantic tasks in response to both continuous and discrete machine-generated prompts, and compare it to the behavior in response to human-generated natural-language prompts. Even when producing a similar output, machine-generated and human prompts trigger different response patterns through the network processing pathways, including different perplexities, different attention and output entropy distributions, and different unit activation profiles. We provide preliminary insight into the nature of the units activated by different prompt types, suggesting that only natural language prompts recruit a genuinely linguistic circuit.", }

Language model prompt optimization research has shown that semantically and grammatically well-formed manually crafted prompts are routinely outperformed by automatically generated token sequences with no apparent meaning or syntactic structure, including sequences of vectors from a model{'}s embedding space. We use machine-generated prompts to probe how models respond to input that is not composed of natural language expressions. We study the behavior of models of different sizes in multiple semantic tasks in response to both continuous and discrete machine-generated prompts, and compare it to the behavior in response to human-generated natural-language prompts. Even when producing a similar output, machine-generated and human prompts trigger different response patterns through the network processing pathways, including different perplexities, different attention and output entropy distributions, and different unit activation profiles. We provide preliminary insight into the nature of the units activated by different prompt types, suggesting that only natural language prompts recruit a genuinely linguistic circuit.

Unnatural language processing: How do language models handle machine-generated prompts?

findings-emnlp.959

2310.15829

Poster

https://aclanthology.org/2023.findings-emnlp.960.bib

https://aclanthology.org/2023.findings-emnlp.960/

@inproceedings{ito-etal-2023-investigating, title = "Investigating the Effectiveness of Multiple Expert Models Collaboration", author = "Ito, Ikumi and Ito, Takumi and Suzuki, Jun and Inui, Kentaro", 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.960", doi = "10.18653/v1/2023.findings-emnlp.960", pages = "14393--14404", abstract = "This paper aims to investigate the effectiveness of several machine translation (MT) models and aggregation methods in a multi-domain setting under fair conditions and explore a direction for tackling multi-domain MT. We mainly compare the performance of the single model approach by jointly training all domains and the multi-expert models approach with a particular aggregation strategy. We conduct experiments on multiple domain datasets and demonstrate that a combination of smaller domain expert models can outperform a larger model trained for all domain data.", }

This paper aims to investigate the effectiveness of several machine translation (MT) models and aggregation methods in a multi-domain setting under fair conditions and explore a direction for tackling multi-domain MT. We mainly compare the performance of the single model approach by jointly training all domains and the multi-expert models approach with a particular aggregation strategy. We conduct experiments on multiple domain datasets and demonstrate that a combination of smaller domain expert models can outperform a larger model trained for all domain data.

Investigating the Effectiveness of Multiple Expert Models Collaboration

findings-emnlp.960

Poster

https://aclanthology.org/2023.findings-emnlp.961.bib

https://aclanthology.org/2023.findings-emnlp.961/

@inproceedings{liu-etal-2023-gradually, title = "Gradually Excavating External Knowledge for Implicit Complex Question Answering", author = "Liu, Chang and Li, Xiaoguang and Shang, Lifeng and Jiang, Xin and Liu, Qun and Lam, Edmund and Wong, Ngai", 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.961", doi = "10.18653/v1/2023.findings-emnlp.961", pages = "14405--14417", abstract = "Recently, large language models (LLMs) have gained much attention for the emergence of human-comparable capabilities and huge potential. However, for open-domain implicit question-answering problems, LLMs may not be the ultimate solution due to the reasons of: 1) uncovered or out-of-date domain knowledge, 2) one-shot generation and hence restricted comprehensiveness. To this end, this work proposes a gradual knowledge excavation framework for open-domain complex question answering, where LLMs iteratively and actively acquire extrinsic information, then reason based on acquired historical knowledge. Specifically, during each step of the solving process, the model selects an action to execute, such as querying external knowledge or performing a single logical reasoning step, to gradually progress toward a final answer. Our method can effectively leverage plug-and-play external knowledge and dynamically adjust the strategy for solving complex questions. Evaluated on the StrategyQA dataset, our method achieves 78.17{\%} accuracy with less than 6{\%} parameters of its competitors, setting new SOTA in the {\textasciitilde}10B LLM class.", }

Recently, large language models (LLMs) have gained much attention for the emergence of human-comparable capabilities and huge potential. However, for open-domain implicit question-answering problems, LLMs may not be the ultimate solution due to the reasons of: 1) uncovered or out-of-date domain knowledge, 2) one-shot generation and hence restricted comprehensiveness. To this end, this work proposes a gradual knowledge excavation framework for open-domain complex question answering, where LLMs iteratively and actively acquire extrinsic information, then reason based on acquired historical knowledge. Specifically, during each step of the solving process, the model selects an action to execute, such as querying external knowledge or performing a single logical reasoning step, to gradually progress toward a final answer. Our method can effectively leverage plug-and-play external knowledge and dynamically adjust the strategy for solving complex questions. Evaluated on the StrategyQA dataset, our method achieves 78.17{\%} accuracy with less than 6{\%} parameters of its competitors, setting new SOTA in the {\textasciitilde}10B LLM class.

Gradually Excavating External Knowledge for Implicit Complex Question Answering

findings-emnlp.961

Poster

https://aclanthology.org/2023.findings-emnlp.962.bib

https://aclanthology.org/2023.findings-emnlp.962/

@inproceedings{zhan-etal-2023-evaluating, title = "Evaluating Subjective Cognitive Appraisals of Emotions from Large Language Models", author = "Zhan, Hongli and Ong, Desmond and Li, Junyi Jessy", 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.962", doi = "10.18653/v1/2023.findings-emnlp.962", pages = "14418--14446", abstract = "The emotions we experience involve complex processes; besides physiological aspects, research in psychology has studied cognitive appraisals where people assess their situations subjectively, according to their own values (Scherer, 2005). Thus, the same situation can often result in different emotional experiences. While the detection of emotion is a well-established task, there is very limited work so far on the automatic prediction of cognitive appraisals. This work fills the gap by presenting CovidET-Appraisals, the most comprehensive dataset to-date that assesses 24 appraisal dimensions, each with a natural language rationale, across 241 Reddit posts. CovidET-Appraisals presents an ideal testbed to evaluate the ability of large language models {---} excelling at a wide range of NLP tasks {---} to automatically assess and explain cognitive appraisals. We found that while the best models are performant, open-sourced LLMs fall short at this task, presenting a new challenge in the future development of emotionally intelligent models. We release our dataset at https://github.com/honglizhan/CovidET-Appraisals-Public.", }

The emotions we experience involve complex processes; besides physiological aspects, research in psychology has studied cognitive appraisals where people assess their situations subjectively, according to their own values (Scherer, 2005). Thus, the same situation can often result in different emotional experiences. While the detection of emotion is a well-established task, there is very limited work so far on the automatic prediction of cognitive appraisals. This work fills the gap by presenting CovidET-Appraisals, the most comprehensive dataset to-date that assesses 24 appraisal dimensions, each with a natural language rationale, across 241 Reddit posts. CovidET-Appraisals presents an ideal testbed to evaluate the ability of large language models {---} excelling at a wide range of NLP tasks {---} to automatically assess and explain cognitive appraisals. We found that while the best models are performant, open-sourced LLMs fall short at this task, presenting a new challenge in the future development of emotionally intelligent models. We release our dataset at https://github.com/honglizhan/CovidET-Appraisals-Public.

Evaluating Subjective Cognitive Appraisals of Emotions from Large Language Models

findings-emnlp.962

2310.14389

Poster

https://aclanthology.org/2023.findings-emnlp.963.bib

https://aclanthology.org/2023.findings-emnlp.963/

@inproceedings{ruzzetti-etal-2023-exploring, title = "Exploring Linguistic Properties of Monolingual {BERT}s with Typological Classification among Languages", author = "Ruzzetti, Elena Sofia and Ranaldi, Federico and Logozzo, Felicia and Mastromattei, Michele and Ranaldi, Leonardo and Zanzotto, Fabio Massimo", 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.963", doi = "10.18653/v1/2023.findings-emnlp.963", pages = "14447--14461", abstract = "The impressive achievements of transformers force NLP researchers to delve into how these models represent the underlying structure of natural language. In this paper, we propose a novel standpoint to investigate the above issue: using typological similarities among languages to observe how their respective monolingual models encode structural information. We aim to layer-wise compare transformers for typologically similar languages to observe whether these similarities emerge for particular layers. For this investigation, we propose to use Centered Kernel Alignment to measure similarity among weight matrices. We found that syntactic typological similarity is consistent with the similarity between the weights in the middle layers, which are the pretrained BERT layers to which syntax encoding is generally attributed. Moreover, we observe that a domain adaptation on semantically equivalent texts enhances this similarity among weight matrices.", }

The impressive achievements of transformers force NLP researchers to delve into how these models represent the underlying structure of natural language. In this paper, we propose a novel standpoint to investigate the above issue: using typological similarities among languages to observe how their respective monolingual models encode structural information. We aim to layer-wise compare transformers for typologically similar languages to observe whether these similarities emerge for particular layers. For this investigation, we propose to use Centered Kernel Alignment to measure similarity among weight matrices. We found that syntactic typological similarity is consistent with the similarity between the weights in the middle layers, which are the pretrained BERT layers to which syntax encoding is generally attributed. Moreover, we observe that a domain adaptation on semantically equivalent texts enhances this similarity among weight matrices.

Exploring Linguistic Properties of Monolingual BERTs with Typological Classification among Languages

findings-emnlp.963

2305.02215

Poster

https://aclanthology.org/2023.findings-emnlp.964.bib

https://aclanthology.org/2023.findings-emnlp.964/

@inproceedings{knaebel-stede-2023-discourse, title = "Discourse Sense Flows: Modelling the Rhetorical Style of Documents across Various Domains", author = "Knaebel, Rene and Stede, Manfred", 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.964", doi = "10.18653/v1/2023.findings-emnlp.964", pages = "14462--14482", abstract = "Recent research on shallow discourse parsing has given renewed attention to the role of discourse relation signals, in particular explicit connectives and so-called alternative lexicalizations. In our work, we first develop new models for extracting signals and classifying their senses, both for explicit connectives and alternative lexicalizations, based on the Penn Discourse Treebank v3 corpus. Thereafter, we apply these models to various raw corpora, and we introduce {`}discourse sense flows{'}, a new way of modeling the rhetorical style of a document by the linear order of coherence relations, as captured by the PDTB senses. The corpora span several genres and domains, and we undertake comparative analyses of the sense flows, as well as experiments on automatic genre/domain discrimination using discourse sense flow patterns as features. We find that n-gram patterns are indeed stronger predictors than simple sense (unigram) distributions.", }

Recent research on shallow discourse parsing has given renewed attention to the role of discourse relation signals, in particular explicit connectives and so-called alternative lexicalizations. In our work, we first develop new models for extracting signals and classifying their senses, both for explicit connectives and alternative lexicalizations, based on the Penn Discourse Treebank v3 corpus. Thereafter, we apply these models to various raw corpora, and we introduce {`}discourse sense flows{'}, a new way of modeling the rhetorical style of a document by the linear order of coherence relations, as captured by the PDTB senses. The corpora span several genres and domains, and we undertake comparative analyses of the sense flows, as well as experiments on automatic genre/domain discrimination using discourse sense flow patterns as features. We find that n-gram patterns are indeed stronger predictors than simple sense (unigram) distributions.

Discourse Sense Flows: Modelling the Rhetorical Style of Documents across Various Domains

findings-emnlp.964

Poster

https://aclanthology.org/2023.findings-emnlp.965.bib

https://aclanthology.org/2023.findings-emnlp.965/

@inproceedings{zhang-zhang-2023-hierarchicalcontrast, title = "{H}ierarchical{C}ontrast: A Coarse-to-Fine Contrastive Learning Framework for Cross-Domain Zero-Shot Slot Filling", author = "Zhang, Junwen and Zhang, Yin", 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.965", doi = "10.18653/v1/2023.findings-emnlp.965", pages = "14483--14503", abstract = "In task-oriented dialogue scenarios, cross-domain zero-shot slot filling plays a vital role in leveraging source domain knowledge to learn a model with high generalization ability in unknown target domain where annotated data is unavailable. However, the existing state-of-the-art zero-shot slot filling methods have limited generalization ability in target domain, they only show effective knowledge transfer on seen slots and perform poorly on unseen slots. To alleviate this issue, we present a novel Hierarchical Contrastive Learning Framework (HiCL) for zero-shot slot filling. Specifically, we propose a coarse- to fine-grained contrastive learning based on Gaussian-distributed embedding to learn the generalized deep semantic relations between utterance-tokens, by optimizing inter- and intra-token distribution distance. This encourages HiCL to generalize to the slot types unseen at training phase. Furthermore, we present a new iterative label set semantics inference method to unbiasedly and separately evaluate the performance of unseen slot types which entangled with their counterparts (i.e., seen slot types) in the previous zero-shot slot filling evaluation methods. The extensive empirical experiments on four datasets demonstrate that the proposed method achieves comparable or even better performance than the current state-of-the-art zero-shot slot filling approaches.", }

In task-oriented dialogue scenarios, cross-domain zero-shot slot filling plays a vital role in leveraging source domain knowledge to learn a model with high generalization ability in unknown target domain where annotated data is unavailable. However, the existing state-of-the-art zero-shot slot filling methods have limited generalization ability in target domain, they only show effective knowledge transfer on seen slots and perform poorly on unseen slots. To alleviate this issue, we present a novel Hierarchical Contrastive Learning Framework (HiCL) for zero-shot slot filling. Specifically, we propose a coarse- to fine-grained contrastive learning based on Gaussian-distributed embedding to learn the generalized deep semantic relations between utterance-tokens, by optimizing inter- and intra-token distribution distance. This encourages HiCL to generalize to the slot types unseen at training phase. Furthermore, we present a new iterative label set semantics inference method to unbiasedly and separately evaluate the performance of unseen slot types which entangled with their counterparts (i.e., seen slot types) in the previous zero-shot slot filling evaluation methods. The extensive empirical experiments on four datasets demonstrate that the proposed method achieves comparable or even better performance than the current state-of-the-art zero-shot slot filling approaches.

HierarchicalContrast: A Coarse-to-Fine Contrastive Learning Framework for Cross-Domain Zero-Shot Slot Filling

findings-emnlp.965

2310.09135

Poster

https://aclanthology.org/2023.findings-emnlp.966.bib

https://aclanthology.org/2023.findings-emnlp.966/

@inproceedings{gomez-rodriguez-williams-2023-confederacy, title = "A Confederacy of Models: a Comprehensive Evaluation of {LLM}s on Creative Writing", author = "G{\'o}mez-Rodr{\'\i}guez, Carlos and Williams, Paul", 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.966", doi = "10.18653/v1/2023.findings-emnlp.966", pages = "14504--14528", abstract = "We evaluate a range of recent LLMs on English creative writing, a challenging and complex task that requires imagination, coherence, and style. We use a difficult, open-ended scenario chosen to avoid training data reuse: an epic narration of a single combat between Ignatius J. Reilly, the protagonist of the Pulitzer Prize-winning novel A Confederacy of Dunces (1980), and a pterodactyl, a prehistoric flying reptile. We ask several LLMs and humans to write such a story and conduct a human evalution involving various criteria such as fluency, coherence, originality, humor, and style. Our results show that some state-of-the-art commercial LLMs match or slightly outperform our writers in most dimensions; whereas open-source LLMs lag behind. Humans retain an edge in creativity, while humor shows a binary divide between LLMs that can handle it comparably to humans and those that fail at it. We discuss the implications and limitations of our study and suggest directions for future research.", }

We evaluate a range of recent LLMs on English creative writing, a challenging and complex task that requires imagination, coherence, and style. We use a difficult, open-ended scenario chosen to avoid training data reuse: an epic narration of a single combat between Ignatius J. Reilly, the protagonist of the Pulitzer Prize-winning novel A Confederacy of Dunces (1980), and a pterodactyl, a prehistoric flying reptile. We ask several LLMs and humans to write such a story and conduct a human evalution involving various criteria such as fluency, coherence, originality, humor, and style. Our results show that some state-of-the-art commercial LLMs match or slightly outperform our writers in most dimensions; whereas open-source LLMs lag behind. Humans retain an edge in creativity, while humor shows a binary divide between LLMs that can handle it comparably to humans and those that fail at it. We discuss the implications and limitations of our study and suggest directions for future research.

A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writing

findings-emnlp.966

2310.08433

Poster

https://aclanthology.org/2023.findings-emnlp.967.bib

https://aclanthology.org/2023.findings-emnlp.967/

@inproceedings{jain-etal-2023-1, title = "1-{PAGER}: One Pass Answer Generation and Evidence Retrieval", author = "Jain, Palak and Soares, Livio and Kwiatkowski, Tom", 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.967", doi = "10.18653/v1/2023.findings-emnlp.967", pages = "14529--14543", abstract = "We present 1-Pager the first system that answers a question and retrieves evidence using a single Transformer-based model and decoding process. 1-Pager incrementally partitions the retrieval corpus using constrained decoding to select a document and answer string, and we show that this is competitive with comparable retrieve-and-read alternatives according to both retrieval and answer accuracy metrics. 1-Pager also outperforms the equivalent {`}closed-book{'} question answering model, by grounding predictions in an evidence corpus. While 1-Pager is not yet on-par with more expensive systems that read many more documents before generating an answer, we argue that it provides an important step toward attributed generation by folding retrieval into the sequence-to-sequence paradigm that is currently dominant in NLP. We also show that the search paths used to partition the corpus are easy to read and understand, paving a way forward for interpretable neural retrieval.", }

We present 1-Pager the first system that answers a question and retrieves evidence using a single Transformer-based model and decoding process. 1-Pager incrementally partitions the retrieval corpus using constrained decoding to select a document and answer string, and we show that this is competitive with comparable retrieve-and-read alternatives according to both retrieval and answer accuracy metrics. 1-Pager also outperforms the equivalent {`}closed-book{'} question answering model, by grounding predictions in an evidence corpus. While 1-Pager is not yet on-par with more expensive systems that read many more documents before generating an answer, we argue that it provides an important step toward attributed generation by folding retrieval into the sequence-to-sequence paradigm that is currently dominant in NLP. We also show that the search paths used to partition the corpus are easy to read and understand, paving a way forward for interpretable neural retrieval.

1-PAGER: One Pass Answer Generation and Evidence Retrieval

findings-emnlp.967

2310.16568

https://huggingface.co/papers/2310.16568

Poster

https://aclanthology.org/2023.findings-emnlp.968.bib

https://aclanthology.org/2023.findings-emnlp.968/

@inproceedings{zhou-etal-2023-context, title = "Context-faithful Prompting for Large Language Models", author = "Zhou, Wenxuan and Zhang, Sheng and Poon, Hoifung and Chen, Muhao", 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.968", doi = "10.18653/v1/2023.findings-emnlp.968", pages = "14544--14556", abstract = "Large language models (LLMs) encode parametric knowledge about world facts and have shown remarkable performance in knowledge-driven NLP tasks. However, their reliance on parametric knowledge may cause them to overlook contextual cues, leading to incorrect predictions in context-sensitive NLP tasks (e.g., knowledge acquisition tasks). In this paper, we seek to assess and enhance LLMs{'} contextual faithfulness in two aspects: knowledge conflict and prediction with abstention. We demonstrate that LLMs{'} faithfulness can be significantly improved using carefully designed prompting strategies. In particular, we identify opinion-based prompts and counterfactual demonstrations as the most effective methods. Opinion-based prompts reframe the context as a narrator{'}s statement and inquire about the narrator{'}s opinions, while counterfactual demonstrations use instances containing false facts to improve faithfulness in knowledge conflict situations. Neither technique requires additional training. We conduct experiments on three datasets of two standard NLP tasks, machine reading comprehension and relation extraction, and the results demonstrate significant improvement in faithfulness to contexts. Code and data are released at https://github.com/wzhouad/context-faithful-llm.", }

Large language models (LLMs) encode parametric knowledge about world facts and have shown remarkable performance in knowledge-driven NLP tasks. However, their reliance on parametric knowledge may cause them to overlook contextual cues, leading to incorrect predictions in context-sensitive NLP tasks (e.g., knowledge acquisition tasks). In this paper, we seek to assess and enhance LLMs{'} contextual faithfulness in two aspects: knowledge conflict and prediction with abstention. We demonstrate that LLMs{'} faithfulness can be significantly improved using carefully designed prompting strategies. In particular, we identify opinion-based prompts and counterfactual demonstrations as the most effective methods. Opinion-based prompts reframe the context as a narrator{'}s statement and inquire about the narrator{'}s opinions, while counterfactual demonstrations use instances containing false facts to improve faithfulness in knowledge conflict situations. Neither technique requires additional training. We conduct experiments on three datasets of two standard NLP tasks, machine reading comprehension and relation extraction, and the results demonstrate significant improvement in faithfulness to contexts. Code and data are released at https://github.com/wzhouad/context-faithful-llm.

Context-faithful Prompting for Large Language Models

findings-emnlp.968

2303.11315

Poster

https://aclanthology.org/2023.findings-emnlp.969.bib

https://aclanthology.org/2023.findings-emnlp.969/

@inproceedings{song-etal-2023-infocl, title = "{I}nfo{CL}: Alleviating Catastrophic Forgetting in Continual Text Classification from An Information Theoretic Perspective", author = "Song, Yifan and Wang, Peiyi and Xiong, Weimin and Zhu, Dawei and Liu, Tianyu and Sui, Zhifang and Li, Sujian", 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.969", doi = "10.18653/v1/2023.findings-emnlp.969", pages = "14557--14570", abstract = "Continual learning (CL) aims to constantly learn new knowledge over time while avoiding catastrophic forgetting on old tasks. We focus on continual text classification under the class-incremental setting. Recent CL studies have identified the severe performance decrease on analogous classes as a key factor for catastrophic forgetting. In this paper, through an in-depth exploration of the representation learning process in CL, we discover that the compression effect of the information bottleneck leads to confusion on analogous classes. To enable the model learn more sufficient representations, we propose a novel replay-based continual text classification method, InfoCL. Our approach utilizes fast-slow and current-past contrastive learning to perform mutual information maximization and better recover the previously learned representations. In addition, InfoCL incorporates an adversarial memory augmentation strategy to alleviate the overfitting problem of replay. Experimental results demonstrate that InfoCL effectively mitigates forgetting and achieves state-of-the-art performance on three text classification tasks.", }

Continual learning (CL) aims to constantly learn new knowledge over time while avoiding catastrophic forgetting on old tasks. We focus on continual text classification under the class-incremental setting. Recent CL studies have identified the severe performance decrease on analogous classes as a key factor for catastrophic forgetting. In this paper, through an in-depth exploration of the representation learning process in CL, we discover that the compression effect of the information bottleneck leads to confusion on analogous classes. To enable the model learn more sufficient representations, we propose a novel replay-based continual text classification method, InfoCL. Our approach utilizes fast-slow and current-past contrastive learning to perform mutual information maximization and better recover the previously learned representations. In addition, InfoCL incorporates an adversarial memory augmentation strategy to alleviate the overfitting problem of replay. Experimental results demonstrate that InfoCL effectively mitigates forgetting and achieves state-of-the-art performance on three text classification tasks.

InfoCL: Alleviating Catastrophic Forgetting in Continual Text Classification from An Information Theoretic Perspective

findings-emnlp.969

2310.06362

Poster

https://aclanthology.org/2023.findings-emnlp.970.bib

https://aclanthology.org/2023.findings-emnlp.970/

@inproceedings{yu-etal-2023-sparse-frame, title = "Sparse Frame Grouping Network with Action Centered for Untrimmed Video Paragraph Captioning", author = "Yu, Guorui and Hu, Yimin and Zhang, Yuejie and Feng, Rui and Zhang, Tao and Gao, Shang", 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.970", doi = "10.18653/v1/2023.findings-emnlp.970", pages = "14571--14580", abstract = "Generating paragraph captions for untrimmed videos without event annotations is challenging, especially when aiming to enhance precision and minimize repetition at the same time. To address this challenge, we propose a module called Sparse Frame Grouping (SFG). It dynamically groups event information with the help of action information for the entire video and excludes redundant frames within pre-defined clips. To enhance the performance, an Intra Contrastive Learning technique is designed to align the SFG module with the core event content in the paragraph, and an Inter Contrastive Learning technique is employed to learn action-guided context with reduced static noise simultaneously. Extensive experiments are conducted on two benchmark datasets (ActivityNet Captions and YouCook2). Results demonstrate that SFG outperforms the state-of-the-art methods on all metrics.", }

Generating paragraph captions for untrimmed videos without event annotations is challenging, especially when aiming to enhance precision and minimize repetition at the same time. To address this challenge, we propose a module called Sparse Frame Grouping (SFG). It dynamically groups event information with the help of action information for the entire video and excludes redundant frames within pre-defined clips. To enhance the performance, an Intra Contrastive Learning technique is designed to align the SFG module with the core event content in the paragraph, and an Inter Contrastive Learning technique is employed to learn action-guided context with reduced static noise simultaneously. Extensive experiments are conducted on two benchmark datasets (ActivityNet Captions and YouCook2). Results demonstrate that SFG outperforms the state-of-the-art methods on all metrics.

Sparse Frame Grouping Network with Action Centered for Untrimmed Video Paragraph Captioning

findings-emnlp.970

Poster

https://aclanthology.org/2023.findings-emnlp.971.bib

https://aclanthology.org/2023.findings-emnlp.971/

@inproceedings{ahmad-luo-2023-unsupervised, title = "Unsupervised Binary Code Translation with Application to Code Clone Detection and Vulnerability Discovery", author = "Ahmad, Iftakhar and Luo, Lannan", 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.971", doi = "10.18653/v1/2023.findings-emnlp.971", pages = "14581--14592", abstract = "Binary code analysis has immense importance in the research domain of software security. Today, software is very often compiled for various Instruction Set Architectures (ISAs). As a result, cross-architecture binary code analysis has become an emerging problem. Recently, deep learning-based binary analysis has shown promising success. It is widely known that training a deep learning model requires a massive amount of data. However, for some low-resource ISAs, an adequate amount of data is hard to find, preventing deep learning from being widely adopted for binary analysis. To overcome the data scarcity problem and facilitate cross-architecture binary code analysis, we propose to apply the ideas and techniques in Neural Machine Translation (NMT) to binary code analysis. Our insight is that a binary, after disassembly, is represented in some assembly language. Given a binary in a low-resource ISA, we translate it to a binary in a high-resource ISA (e.g., x86). Then we can use a model that has been trained on the high-resource ISA to test the translated binary. We have implemented the model called UNSUPERBINTRANS, and conducted experiments to evaluate its performance. Specifically, we conducted two downstream tasks, including code similarity detection and vulnerability discovery. In both tasks, we achieved high accuracies.", }

Binary code analysis has immense importance in the research domain of software security. Today, software is very often compiled for various Instruction Set Architectures (ISAs). As a result, cross-architecture binary code analysis has become an emerging problem. Recently, deep learning-based binary analysis has shown promising success. It is widely known that training a deep learning model requires a massive amount of data. However, for some low-resource ISAs, an adequate amount of data is hard to find, preventing deep learning from being widely adopted for binary analysis. To overcome the data scarcity problem and facilitate cross-architecture binary code analysis, we propose to apply the ideas and techniques in Neural Machine Translation (NMT) to binary code analysis. Our insight is that a binary, after disassembly, is represented in some assembly language. Given a binary in a low-resource ISA, we translate it to a binary in a high-resource ISA (e.g., x86). Then we can use a model that has been trained on the high-resource ISA to test the translated binary. We have implemented the model called UNSUPERBINTRANS, and conducted experiments to evaluate its performance. Specifically, we conducted two downstream tasks, including code similarity detection and vulnerability discovery. In both tasks, we achieved high accuracies.

Unsupervised Binary Code Translation with Application to Code Clone Detection and Vulnerability Discovery

findings-emnlp.971

Poster

https://aclanthology.org/2023.findings-emnlp.972.bib

https://aclanthology.org/2023.findings-emnlp.972/

@inproceedings{nair-etal-2023-drilling, title = "Drilling Down into the Discourse Structure with {LLM}s for Long Document Question Answering", author = "Nair, Inderjeet and Somasundaram, Shwetha and Saxena, Apoorv and Goswami, Koustava", 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.972", doi = "10.18653/v1/2023.findings-emnlp.972", pages = "14593--14606", abstract = "We address the task of evidence retrieval for long document question answering, which involves locating relevant paragraphs within a document to answer a question. We aim to assess the applicability of large language models (LLMs) in the task of zero-shot long document evidence retrieval, owing to their unprecedented performance across various NLP tasks. However, currently the LLMs can consume limited context lengths as input, thus providing document chunks as inputs might overlook the global context while missing out on capturing the inter-segment dependencies. Moreover, directly feeding the large input sets can incur significant computational costs, particularly when processing the entire document (and potentially incurring monetary expenses with enterprise APIs like OpenAI{'}s GPT variants). To address these challenges, we propose a suite of techniques that exploit the discourse structure commonly found in documents. By utilizing this structure, we create a condensed representation of the document, enabling a more comprehensive understanding and analysis of relationships between different parts. We retain 99.6{\%} of the best zero-shot approach{'}s performance, while processing only 26{\%} of the total tokens used by the best approach in the information seeking evidence retrieval setup. We also show how our approach can be combined with *self-ask* reasoning agent to achieve best zero-shot performance in complex multi-hop question answering, just $\approx 4${\%} short of zero-shot performance using gold evidence.", }

We address the task of evidence retrieval for long document question answering, which involves locating relevant paragraphs within a document to answer a question. We aim to assess the applicability of large language models (LLMs) in the task of zero-shot long document evidence retrieval, owing to their unprecedented performance across various NLP tasks. However, currently the LLMs can consume limited context lengths as input, thus providing document chunks as inputs might overlook the global context while missing out on capturing the inter-segment dependencies. Moreover, directly feeding the large input sets can incur significant computational costs, particularly when processing the entire document (and potentially incurring monetary expenses with enterprise APIs like OpenAI{'}s GPT variants). To address these challenges, we propose a suite of techniques that exploit the discourse structure commonly found in documents. By utilizing this structure, we create a condensed representation of the document, enabling a more comprehensive understanding and analysis of relationships between different parts. We retain 99.6{\%} of the best zero-shot approach{'}s performance, while processing only 26{\%} of the total tokens used by the best approach in the information seeking evidence retrieval setup. We also show how our approach can be combined with *self-ask* reasoning agent to achieve best zero-shot performance in complex multi-hop question answering, just $\approx 4${\%} short of zero-shot performance using gold evidence.

Drilling Down into the Discourse Structure with LLMs for Long Document Question Answering

findings-emnlp.972

2311.13565

Poster

https://aclanthology.org/2023.findings-emnlp.973.bib

https://aclanthology.org/2023.findings-emnlp.973/

@inproceedings{michaelov-bergen-2023-emergent, title = "Emergent Inabilities? Inverse Scaling Over the Course of Pretraining", author = "Michaelov, James and Bergen, Ben", 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.973", doi = "10.18653/v1/2023.findings-emnlp.973", pages = "14607--14615", abstract = "Does inverse scaling only occur as a function of model size, or can it also occur over the course of training? We carry out an exploratory study investigating whether the performance of language models on specific tasks can decrease (while general performance remains high) during training on the language modeling task. We find 8 tasks on which Pythia 12B (Biderman et al., 2023) shows decreased performance over the course of training. Five of these tasks (TruthfulQA-MC1, TruthfulQA-MC2, Hindsight Neglect, Memo Trap, and Pattern Match Suppression) additionally show a consistent relationship whereby larger language models show a greater decrease in performance the more they are trained, despite showing standard (positive) scaling overall. This highlights the importance of testing performance at all relevant benchmarks any time models are trained on additional data, even if their overall performance improves.", }

Does inverse scaling only occur as a function of model size, or can it also occur over the course of training? We carry out an exploratory study investigating whether the performance of language models on specific tasks can decrease (while general performance remains high) during training on the language modeling task. We find 8 tasks on which Pythia 12B (Biderman et al., 2023) shows decreased performance over the course of training. Five of these tasks (TruthfulQA-MC1, TruthfulQA-MC2, Hindsight Neglect, Memo Trap, and Pattern Match Suppression) additionally show a consistent relationship whereby larger language models show a greater decrease in performance the more they are trained, despite showing standard (positive) scaling overall. This highlights the importance of testing performance at all relevant benchmarks any time models are trained on additional data, even if their overall performance improves.

Emergent Inabilities? Inverse Scaling Over the Course of Pretraining

findings-emnlp.973

2305.14681

Poster

https://aclanthology.org/2023.findings-emnlp.974.bib

https://aclanthology.org/2023.findings-emnlp.974/

@inproceedings{jin-etal-2023-alignment, title = "Alignment Precedes Fusion: Open-Vocabulary Named Entity Recognition as Context-Type Semantic Matching", author = "Jin, Zhuoran and Cao, Pengfei and He, Zhitao and Chen, Yubo and Liu, Kang and Zhao, Jun", 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.974", doi = "10.18653/v1/2023.findings-emnlp.974", pages = "14616--14637", abstract = "Despite the significant progress in developing named entity recognition models, scaling to novel-emerging types still remains challenging in real-world scenarios. Continual learning and zero-shot learning approaches have been explored to handle novel-emerging types with less human supervision, but they have not been as successfully adopted as supervised approaches. Meanwhile, humans possess a much larger vocabulary size than these approaches and have the ability to learn the alignment between entities and concepts effortlessly through natural supervision. In this paper, we consider a more realistic and challenging setting called open-vocabulary named entity recognition (OVNER) to imitate human-level ability. OVNER aims to recognize entities in novel types by their textual names or descriptions. Specifically, we formulate OVNER as a semantic matching task and propose a novel and scalable two-stage method called Context-Type SemAntiC Alignment and FusiOn (CACAO). In the pre-training stage, we adopt Dual-Encoder for context-type semantic alignment and pre-train Dual-Encoder on 80M context-type pairs which are easily accessible through natural supervision. In the fine-tuning stage, we use Cross-Encoder for context-type semantic fusion and fine-tune Cross-Encoder on base types with human supervision. Experimental results show that our method outperforms the previous state-of-the-art methods on three challenging OVNER benchmarks by 9.7{\%}, 9.5{\%}, and 1.8{\%} F1-score of novel types. Moreover, CACAO also demonstrates its flexible transfer ability in cross-domain NER.", }

Despite the significant progress in developing named entity recognition models, scaling to novel-emerging types still remains challenging in real-world scenarios. Continual learning and zero-shot learning approaches have been explored to handle novel-emerging types with less human supervision, but they have not been as successfully adopted as supervised approaches. Meanwhile, humans possess a much larger vocabulary size than these approaches and have the ability to learn the alignment between entities and concepts effortlessly through natural supervision. In this paper, we consider a more realistic and challenging setting called open-vocabulary named entity recognition (OVNER) to imitate human-level ability. OVNER aims to recognize entities in novel types by their textual names or descriptions. Specifically, we formulate OVNER as a semantic matching task and propose a novel and scalable two-stage method called Context-Type SemAntiC Alignment and FusiOn (CACAO). In the pre-training stage, we adopt Dual-Encoder for context-type semantic alignment and pre-train Dual-Encoder on 80M context-type pairs which are easily accessible through natural supervision. In the fine-tuning stage, we use Cross-Encoder for context-type semantic fusion and fine-tune Cross-Encoder on base types with human supervision. Experimental results show that our method outperforms the previous state-of-the-art methods on three challenging OVNER benchmarks by 9.7{\%}, 9.5{\%}, and 1.8{\%} F1-score of novel types. Moreover, CACAO also demonstrates its flexible transfer ability in cross-domain NER.

Alignment Precedes Fusion: Open-Vocabulary Named Entity Recognition as Context-Type Semantic Matching

findings-emnlp.974

Poster

https://aclanthology.org/2023.findings-emnlp.975.bib

https://aclanthology.org/2023.findings-emnlp.975/

@inproceedings{razdaibiedina-etal-2023-representation, title = "Representation Projection Invariance Mitigates Representation Collapse", author = "Razdaibiedina, Anastasia and Khetan, Ashish and Karnin, Zohar and Khashabi, Daniel and Madan, Vivek", 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.975", doi = "10.18653/v1/2023.findings-emnlp.975", pages = "14638--14664", abstract = "Fine-tuning contextualized representations learned by pre-trained language models remains a prevalent practice in NLP. However, fine-tuning can lead to representation degradation (also known as representation collapse), which may result in instability, sub-optimal performance, and weak generalization. In this paper, we propose Representation Projection Invariance (REPINA), a novel regularization method to maintain the information content of representation and reduce representation collapse during fine-tuning by discouraging undesirable changes in the representations. We study the empirical behavior of the proposed regularization in comparison to 5 comparable baselines across 13 language understanding tasks (GLUE benchmark and six additional datasets). When evaluating in-domain performance, REPINA consistently outperforms other baselines on most tasks (10 out of 13). Additionally, REPINA improves out-of-distribution performance. We also demonstrate its effectiveness in few-shot settings and robustness to label perturbation. As a by-product, we extend previous studies of representation collapse and propose several metrics to quantify it. Our empirical findings show that our approach is significantly more effective at mitigating representation collapse.", }

Fine-tuning contextualized representations learned by pre-trained language models remains a prevalent practice in NLP. However, fine-tuning can lead to representation degradation (also known as representation collapse), which may result in instability, sub-optimal performance, and weak generalization. In this paper, we propose Representation Projection Invariance (REPINA), a novel regularization method to maintain the information content of representation and reduce representation collapse during fine-tuning by discouraging undesirable changes in the representations. We study the empirical behavior of the proposed regularization in comparison to 5 comparable baselines across 13 language understanding tasks (GLUE benchmark and six additional datasets). When evaluating in-domain performance, REPINA consistently outperforms other baselines on most tasks (10 out of 13). Additionally, REPINA improves out-of-distribution performance. We also demonstrate its effectiveness in few-shot settings and robustness to label perturbation. As a by-product, we extend previous studies of representation collapse and propose several metrics to quantify it. Our empirical findings show that our approach is significantly more effective at mitigating representation collapse.

Representation Projection Invariance Mitigates Representation Collapse

findings-emnlp.975

2205.11603

Poster

https://aclanthology.org/2023.findings-emnlp.976.bib

https://aclanthology.org/2023.findings-emnlp.976/

@inproceedings{dong-etal-2023-tunable, title = "Tunable Soft Prompts are Messengers in Federated Learning", author = "Dong, Chenhe and Xie, Yuexiang and Ding, Bolin and Shen, Ying and Li, Yaliang", 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.976", doi = "10.18653/v1/2023.findings-emnlp.976", pages = "14665--14675", abstract = "Federated learning (FL) enables multiple participants to collaboratively train machine learning models using decentralized data sources, alleviating privacy concerns that arise from directly sharing local data. However, the lack of model privacy protection in FL becomes an unneglectable challenge, especially when people want to federally finetune models based on a proprietary large language model. In this study, we propose a novel FL training approach that accomplishes information exchange among participants via tunable soft prompts. These soft prompts, updated and transmitted between the server and clients, assume the role of the global model parameters and serve as messengers to deliver useful knowledge from the local data and global model. As the global model itself is not required to be shared and the local training is conducted based on an auxiliary model with fewer parameters than the global model, the proposed approach provides protection for the global model while reducing communication and computation costs in FL. Extensive experiments show the effectiveness of the proposed approach compared to several baselines. We have released the source code at https://github.com/alibaba/FederatedScope/tree/fedsp/federatedscope/nlp/fedsp.", }

Federated learning (FL) enables multiple participants to collaboratively train machine learning models using decentralized data sources, alleviating privacy concerns that arise from directly sharing local data. However, the lack of model privacy protection in FL becomes an unneglectable challenge, especially when people want to federally finetune models based on a proprietary large language model. In this study, we propose a novel FL training approach that accomplishes information exchange among participants via tunable soft prompts. These soft prompts, updated and transmitted between the server and clients, assume the role of the global model parameters and serve as messengers to deliver useful knowledge from the local data and global model. As the global model itself is not required to be shared and the local training is conducted based on an auxiliary model with fewer parameters than the global model, the proposed approach provides protection for the global model while reducing communication and computation costs in FL. Extensive experiments show the effectiveness of the proposed approach compared to several baselines. We have released the source code at https://github.com/alibaba/FederatedScope/tree/fedsp/federatedscope/nlp/fedsp.

Tunable Soft Prompts are Messengers in Federated Learning

findings-emnlp.976

2311.06805

Poster

https://aclanthology.org/2023.findings-emnlp.977.bib

https://aclanthology.org/2023.findings-emnlp.977/

@inproceedings{yan-etal-2023-style, title = "Style-Aware Radiology Report Generation with {R}ad{G}raph and Few-Shot Prompting", author = "Yan, Benjamin and Liu, Ruochen and Kuo, David and Adithan, Subathra and Reis, Eduardo and Kwak, Stephen and Venugopal, Vasantha and O{'}Connell, Chloe and Saenz, Agustina and Rajpurkar, Pranav and Moor, Michael", 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.977", doi = "10.18653/v1/2023.findings-emnlp.977", pages = "14676--14688", abstract = "Automatically generated reports from medical images promise to improve the workflow of radiologists. Existing methods consider an image-to-report modeling task by directly generating a fully-fledged report from an image. However, this conflates the content of the report (e.g., findings and their attributes) with its style (e.g., format and choice of words), which can lead to clinically inaccurate reports. To address this, we propose a two-step approach for radiology report generation. First, we extract the content from an image; then, we verbalize the extracted content into a report that matches the style of a specific radiologist. For this, we leverage RadGraph{---}a graph representation of reports{---}together with large language models (LLMs). In our quantitative evaluations, we find that our approach leads to beneficial performance. Our human evaluation with clinical raters highlights that the AI-generated reports are indistinguishably tailored to the style of individual radiologist despite leveraging only a few examples as context.", }

Automatically generated reports from medical images promise to improve the workflow of radiologists. Existing methods consider an image-to-report modeling task by directly generating a fully-fledged report from an image. However, this conflates the content of the report (e.g., findings and their attributes) with its style (e.g., format and choice of words), which can lead to clinically inaccurate reports. To address this, we propose a two-step approach for radiology report generation. First, we extract the content from an image; then, we verbalize the extracted content into a report that matches the style of a specific radiologist. For this, we leverage RadGraph{---}a graph representation of reports{---}together with large language models (LLMs). In our quantitative evaluations, we find that our approach leads to beneficial performance. Our human evaluation with clinical raters highlights that the AI-generated reports are indistinguishably tailored to the style of individual radiologist despite leveraging only a few examples as context.

Style-Aware Radiology Report Generation with RadGraph and Few-Shot Prompting

findings-emnlp.977

2310.17811

Poster

https://aclanthology.org/2023.findings-emnlp.978.bib

https://aclanthology.org/2023.findings-emnlp.978/

@inproceedings{zuo-etal-2023-incorporating, title = "Incorporating Probing Signals into Multimodal Machine Translation via Visual Question-Answering Pairs", author = "Zuo, Yuxin and Li, Bei and Lv, Chuanhao and Zheng, Tong and Xiao, Tong and Zhu, JingBo", 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.978", doi = "10.18653/v1/2023.findings-emnlp.978", pages = "14689--14701", abstract = "This paper presents an in-depth study of multimodal machine translation (MMT), examining the prevailing understanding that MMT systems exhibit decreased sensitivity to visual information when text inputs are complete. Instead, we attribute this phenomenon to insufficient cross-modal interaction, rather than image information redundancy. A novel approach is proposed to generate parallel Visual Question-Answering (VQA) style pairs from the source text, fostering more robust cross-modal interaction. Using Large Language Models (LLMs), we explicitly model the probing signal in MMT to convert it into VQA-style data to create the Multi30K-VQA dataset. An MMT-VQA multitask learning framework is introduced to incorporate explicit probing signals from the dataset into the MMT training process. Experimental results on two widely-used benchmarks demonstrate the effectiveness of this novel approach. Our code and data would be available at: \url{https://github.com/libeineu/MMT-VQA}.", }

This paper presents an in-depth study of multimodal machine translation (MMT), examining the prevailing understanding that MMT systems exhibit decreased sensitivity to visual information when text inputs are complete. Instead, we attribute this phenomenon to insufficient cross-modal interaction, rather than image information redundancy. A novel approach is proposed to generate parallel Visual Question-Answering (VQA) style pairs from the source text, fostering more robust cross-modal interaction. Using Large Language Models (LLMs), we explicitly model the probing signal in MMT to convert it into VQA-style data to create the Multi30K-VQA dataset. An MMT-VQA multitask learning framework is introduced to incorporate explicit probing signals from the dataset into the MMT training process. Experimental results on two widely-used benchmarks demonstrate the effectiveness of this novel approach. Our code and data would be available at: \url{https://github.com/libeineu/MMT-VQA}.

Incorporating Probing Signals into Multimodal Machine Translation via Visual Question-Answering Pairs

findings-emnlp.978

2310.17133

Poster

https://aclanthology.org/2023.findings-emnlp.979.bib

https://aclanthology.org/2023.findings-emnlp.979/

@inproceedings{cao-etal-2023-genkie, title = "{G}en{KIE}: Robust Generative Multimodal Document Key Information Extraction", author = "Cao, Panfeng and Wang, Ye and Zhang, Qiang and Meng, Zaiqiao", 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.979", doi = "10.18653/v1/2023.findings-emnlp.979", pages = "14702--14713", abstract = "Key information extraction (KIE) from scanned documents has gained increasing attention because of its applications in various domains. Although promising results have been achieved by some recent KIE approaches, they are usually built based on discriminative models, which lack the ability to handle optical character recognition (OCR) errors and require laborious token-level labeling. In this paper, we propose a novel generative end-to-end model, named GenKIE, to address the KIE task. GenKIE is a sequence-to-sequence multimodal generative model that utilizes multimodal encoders to embed visual, layout and textual features and a decoder to generate the desired output. Well-designed prompts are leveraged to incorporate the label semantics as the weakly supervised signals and entice the generation of the key information. One notable advantage of the generative model is that it enables automatic correction of OCR errors. Besides, token-level granular annotation is not required. Extensive experiments on multiple public real-world datasets show that GenKIE effectively generalizes over different types of documents and achieves state-of-the-art results. Our experiments also validate the model{'}s robustness against OCR errors, making GenKIE highly applicable in real-world scenarios.", }

Key information extraction (KIE) from scanned documents has gained increasing attention because of its applications in various domains. Although promising results have been achieved by some recent KIE approaches, they are usually built based on discriminative models, which lack the ability to handle optical character recognition (OCR) errors and require laborious token-level labeling. In this paper, we propose a novel generative end-to-end model, named GenKIE, to address the KIE task. GenKIE is a sequence-to-sequence multimodal generative model that utilizes multimodal encoders to embed visual, layout and textual features and a decoder to generate the desired output. Well-designed prompts are leveraged to incorporate the label semantics as the weakly supervised signals and entice the generation of the key information. One notable advantage of the generative model is that it enables automatic correction of OCR errors. Besides, token-level granular annotation is not required. Extensive experiments on multiple public real-world datasets show that GenKIE effectively generalizes over different types of documents and achieves state-of-the-art results. Our experiments also validate the model{'}s robustness against OCR errors, making GenKIE highly applicable in real-world scenarios.

GenKIE: Robust Generative Multimodal Document Key Information Extraction

findings-emnlp.979

2310.16131

Poster

https://aclanthology.org/2023.findings-emnlp.980.bib

https://aclanthology.org/2023.findings-emnlp.980/

@inproceedings{nguyen-etal-2023-improving-multimodal, title = "Improving Multimodal Sentiment Analysis: Supervised Angular margin-based Contrastive Learning for Enhanced Fusion Representation", author = "Nguyen, Cong-Duy and Nguyen, Thong and Vu, Duc and Luu, Anh", 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.980", doi = "10.18653/v1/2023.findings-emnlp.980", pages = "14714--14724", abstract = "The effectiveness of a model is heavily reliant on the quality of the fusion representation of multiple modalities in multimodal sentiment analysis. Moreover, each modality is extracted from raw input and integrated with the rest to construct a multimodal representation. Although previous methods have proposed multimodal representations and achieved promising results, most of them focus on forming positive and negative pairs, neglecting the variation in sentiment scores within the same class. Additionally, they fail to capture the significance of unimodal representations in the fusion vector. To address these limitations, we introduce a framework called Supervised Angular-based Contrastive Learning for Multimodal Sentiment Analysis. This framework aims to enhance discrimination and generalizability of the multimodal representation and overcome biases in the fusion vector{'}s modality. Our experimental results, along with visualizations on two widely used datasets, demonstrate the effectiveness of our approach.", }

The effectiveness of a model is heavily reliant on the quality of the fusion representation of multiple modalities in multimodal sentiment analysis. Moreover, each modality is extracted from raw input and integrated with the rest to construct a multimodal representation. Although previous methods have proposed multimodal representations and achieved promising results, most of them focus on forming positive and negative pairs, neglecting the variation in sentiment scores within the same class. Additionally, they fail to capture the significance of unimodal representations in the fusion vector. To address these limitations, we introduce a framework called Supervised Angular-based Contrastive Learning for Multimodal Sentiment Analysis. This framework aims to enhance discrimination and generalizability of the multimodal representation and overcome biases in the fusion vector{'}s modality. Our experimental results, along with visualizations on two widely used datasets, demonstrate the effectiveness of our approach.

Improving Multimodal Sentiment Analysis: Supervised Angular margin-based Contrastive Learning for Enhanced Fusion Representation

findings-emnlp.980

2312.02227

Poster

https://aclanthology.org/2023.findings-emnlp.981.bib

https://aclanthology.org/2023.findings-emnlp.981/

@inproceedings{ushio-etal-2023-efficient, title = "Efficient Multilingual Language Model Compression through Vocabulary Trimming", author = "Ushio, Asahi and Zhou, Yi and Camacho-Collados, Jose", 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.981", doi = "10.18653/v1/2023.findings-emnlp.981", pages = "14725--14739", abstract = "Multilingual language models (LMs) have become a powerful tool in NLP, especially for non-English languages. Nevertheless, model parameters of multilingual LMs remain large due to the larger embedding matrix of the vocabulary covering tokens in different languages. Instead, monolingual LMs can be trained in a target language with the language-specific vocabulary only. In this paper, we propose vocabulary-trimming (VT), a method to reduce a multilingual LM vocabulary to a target language by deleting potentially irrelevant tokens from its vocabulary. In theory, VT can compress any existing multilingual LM to any language covered by the original model. In our experiments, we show that VT can retain the original performance of the multilingual LM, while being considerably smaller in size than the original multilingual LM. The evaluation is performed over four NLP tasks (two generative and two classification tasks) among four widely used multilingual LMs in seven languages. The results show that this methodology can keep the best of both monolingual and multilingual worlds by keeping a small size as monolingual models without the need for specifically retraining them, and can even help limit potentially harmful social biases.", }

Multilingual language models (LMs) have become a powerful tool in NLP, especially for non-English languages. Nevertheless, model parameters of multilingual LMs remain large due to the larger embedding matrix of the vocabulary covering tokens in different languages. Instead, monolingual LMs can be trained in a target language with the language-specific vocabulary only. In this paper, we propose vocabulary-trimming (VT), a method to reduce a multilingual LM vocabulary to a target language by deleting potentially irrelevant tokens from its vocabulary. In theory, VT can compress any existing multilingual LM to any language covered by the original model. In our experiments, we show that VT can retain the original performance of the multilingual LM, while being considerably smaller in size than the original multilingual LM. The evaluation is performed over four NLP tasks (two generative and two classification tasks) among four widely used multilingual LMs in seven languages. The results show that this methodology can keep the best of both monolingual and multilingual worlds by keeping a small size as monolingual models without the need for specifically retraining them, and can even help limit potentially harmful social biases.

Efficient Multilingual Language Model Compression through Vocabulary Trimming

findings-emnlp.981

https://huggingface.co/papers/2305.15020

Poster

https://aclanthology.org/2023.findings-emnlp.982.bib

https://aclanthology.org/2023.findings-emnlp.982/

@inproceedings{wu-2023-icu, title = "{ICU}: Conquering Language Barriers in Vision-and-Language Modeling by Dividing the Tasks into Image Captioning and Language Understanding", author = "Wu, Guojun", 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.982", doi = "10.18653/v1/2023.findings-emnlp.982", pages = "14740--14746", abstract = "Most multilingual vision-and-language (V{\&}L) research aims to accomplish multilingual and multimodal capabilities within one model. However, the scarcity of multilingual captions for images has hindered the development. To overcome this obstacle, we propose ICU, Image Caption Understanding, which divides a V{\&}L task into two stages: a V{\&}L model performs image captioning in English, and a multilingual language model (mLM), in turn, takes the caption as the alt text and performs cross-lingual language understanding. The burden of multilingual processing is lifted off V{\&}L model and placed on mLM. Since the multilingual text data is relatively of higher abundance and quality, ICU can facilitate the conquering of language barriers for V{\&}L models. In experiments on two tasks across 9 languages in the IGLUE benchmark, we show that ICU can achieve new state-of-the-art results for five languages, and comparable results for the rest.", }

Most multilingual vision-and-language (V{\&}L) research aims to accomplish multilingual and multimodal capabilities within one model. However, the scarcity of multilingual captions for images has hindered the development. To overcome this obstacle, we propose ICU, Image Caption Understanding, which divides a V{\&}L task into two stages: a V{\&}L model performs image captioning in English, and a multilingual language model (mLM), in turn, takes the caption as the alt text and performs cross-lingual language understanding. The burden of multilingual processing is lifted off V{\&}L model and placed on mLM. Since the multilingual text data is relatively of higher abundance and quality, ICU can facilitate the conquering of language barriers for V{\&}L models. In experiments on two tasks across 9 languages in the IGLUE benchmark, we show that ICU can achieve new state-of-the-art results for five languages, and comparable results for the rest.

ICU: Conquering Language Barriers in Vision-and-Language Modeling by Dividing the Tasks into Image Captioning and Language Understanding

findings-emnlp.982

2310.12531

Poster

https://aclanthology.org/2023.findings-emnlp.983.bib

https://aclanthology.org/2023.findings-emnlp.983/

@inproceedings{kim-cho-2023-gta, title = "{GTA}: Gated Toxicity Avoidance for {LM} Performance Preservation", author = "Kim, Heegyu and Cho, Hyunsouk", 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.983", doi = "10.18653/v1/2023.findings-emnlp.983", pages = "14747--14763", abstract = "Caution: This paper includes offensive words that could potentially cause unpleasantness. The fast-paced evolution of generative language models such as GPT-4 has demonstrated outstanding results in various NLP generation tasks. However, due to the potential generation of offensive words related to race or gender, various Controllable Text Generation (CTG) methods have been proposed to mitigate the occurrence of harmful words. However, existing CTG methods not only reduce toxicity but also negatively impact several aspects of the language model{'}s generation performance, including topic consistency, grammar, and perplexity. This paper explores the limitations of previous methods and introduces a novel solution in the form of a simple Gated Toxicity Avoidance (GTA) that can be applied to any CTG method. We also evaluate the effectiveness of the proposed GTA by comparing it with state-of-the-art CTG methods across various datasets. Our findings reveal that gated toxicity avoidance efficiently achieves comparable levels of toxicity reduction to the original CTG methods while preserving the generation performance of the language model.", }

Caution: This paper includes offensive words that could potentially cause unpleasantness. The fast-paced evolution of generative language models such as GPT-4 has demonstrated outstanding results in various NLP generation tasks. However, due to the potential generation of offensive words related to race or gender, various Controllable Text Generation (CTG) methods have been proposed to mitigate the occurrence of harmful words. However, existing CTG methods not only reduce toxicity but also negatively impact several aspects of the language model{'}s generation performance, including topic consistency, grammar, and perplexity. This paper explores the limitations of previous methods and introduces a novel solution in the form of a simple Gated Toxicity Avoidance (GTA) that can be applied to any CTG method. We also evaluate the effectiveness of the proposed GTA by comparing it with state-of-the-art CTG methods across various datasets. Our findings reveal that gated toxicity avoidance efficiently achieves comparable levels of toxicity reduction to the original CTG methods while preserving the generation performance of the language model.

GTA: Gated Toxicity Avoidance for LM Performance Preservation

findings-emnlp.983

2312.06122

https://huggingface.co/papers/2312.06122

Poster

https://aclanthology.org/2023.findings-emnlp.984.bib

https://aclanthology.org/2023.findings-emnlp.984/

@inproceedings{xu-etal-2023-lmgqs, title = "{LMGQS}: A Large-scale Dataset for Query-focused Summarization", author = "Xu, Ruochen and Wang, Song and Liu, Yang and Wang, Shuohang and Xu, Yichong and Iter, Dan and He, Pengcheng and Zhu, Chenguang and Zeng, Michael", 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.984", doi = "10.18653/v1/2023.findings-emnlp.984", pages = "14764--14776", abstract = "Query-focused summarization (QFS) aims to extract or generate a summary of an input document that directly answers or is relevant to a given query. The lack of large-scale datasets in the form of documents, queries, and summaries has hindered model development in this area. In contrast, multiple large-scale high-quality datasets for generic summarization exist. We hypothesize that there is a hidden query for each summary sentence in a generic summarization annotation, and we utilize a large-scale pretrained language model to recover it. In this way, we convert four generic summarization benchmarks into a new QFS benchmark dataset, LMGQS, which consists of over 1 million document-query-summary samples. We thoroughly investigate the properties of our proposed dataset and establish baselines with state-of-the-art summarization models. By fine-tuning a language model on LMGQS, we achieve state-of-the-art zero-shot and supervised performance on multiple existing QFS benchmarks, demonstrating the high quality and diversity of LMGQS.", }

Query-focused summarization (QFS) aims to extract or generate a summary of an input document that directly answers or is relevant to a given query. The lack of large-scale datasets in the form of documents, queries, and summaries has hindered model development in this area. In contrast, multiple large-scale high-quality datasets for generic summarization exist. We hypothesize that there is a hidden query for each summary sentence in a generic summarization annotation, and we utilize a large-scale pretrained language model to recover it. In this way, we convert four generic summarization benchmarks into a new QFS benchmark dataset, LMGQS, which consists of over 1 million document-query-summary samples. We thoroughly investigate the properties of our proposed dataset and establish baselines with state-of-the-art summarization models. By fine-tuning a language model on LMGQS, we achieve state-of-the-art zero-shot and supervised performance on multiple existing QFS benchmarks, demonstrating the high quality and diversity of LMGQS.

LMGQS: A Large-scale Dataset for Query-focused Summarization

findings-emnlp.984

2305.13086

Poster

https://aclanthology.org/2023.findings-emnlp.985.bib

https://aclanthology.org/2023.findings-emnlp.985/

@inproceedings{chen-etal-2023-chatcot, title = "{C}hat{C}o{T}: Tool-Augmented Chain-of-Thought Reasoning on Chat-based Large Language Models", author = "Chen, Zhipeng and Zhou, Kun and Zhang, Beichen and Gong, Zheng and Zhao, Xin 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.985", doi = "10.18653/v1/2023.findings-emnlp.985", pages = "14777--14790", abstract = "Although large language models (LLMs) have achieved excellent performance in a variety of evaluation benchmarks, they still struggle in complex reasoning tasks which require specific knowledge and multi-hop reasoning. To improve the reasoning abilities, we propose $\textbf{ChatCoT}$, a tool-augmented chain-of-thought reasoning framework for chat-based LLMs ($\textit{e.g.,}$ ChatGPT). In ChatCoT, we model the chain-of-thought (CoT) reasoning as multi-turn conversations, to utilize tools in a more natural way through chatting. At each turn, LLMs can either interact with tools or perform the reasoning. Our approach can effectively leverage the multi-turn conversation ability of chat-based LLMs, and integrate the thought chain following and tools manipulation in a unified way. Specially, we initialize the early turns of the conversation by the knowledge about tools, tasks, and reasoning format, and propose an iterative $\textit{tool-augmented reasoning}$ step to perform step-by-step tool-augmented reasoning. The experiment results on two complex reasoning datasets (MATH and HotpotQA) have shown the effectiveness of ChatCoT on complex reasoning tasks, achieving a 7.9{\%} relative improvement over the state-of-the-art baseline.", }

Although large language models (LLMs) have achieved excellent performance in a variety of evaluation benchmarks, they still struggle in complex reasoning tasks which require specific knowledge and multi-hop reasoning. To improve the reasoning abilities, we propose $\textbf{ChatCoT}$, a tool-augmented chain-of-thought reasoning framework for chat-based LLMs ($\textit{e.g.,}$ ChatGPT). In ChatCoT, we model the chain-of-thought (CoT) reasoning as multi-turn conversations, to utilize tools in a more natural way through chatting. At each turn, LLMs can either interact with tools or perform the reasoning. Our approach can effectively leverage the multi-turn conversation ability of chat-based LLMs, and integrate the thought chain following and tools manipulation in a unified way. Specially, we initialize the early turns of the conversation by the knowledge about tools, tasks, and reasoning format, and propose an iterative $\textit{tool-augmented reasoning}$ step to perform step-by-step tool-augmented reasoning. The experiment results on two complex reasoning datasets (MATH and HotpotQA) have shown the effectiveness of ChatCoT on complex reasoning tasks, achieving a 7.9{\%} relative improvement over the state-of-the-art baseline.

ChatCoT: Tool-Augmented Chain-of-Thought Reasoning on Chat-based Large Language Models

findings-emnlp.985

2305.14323

https://huggingface.co/papers/2305.14323

Poster

https://aclanthology.org/2023.findings-emnlp.986.bib

https://aclanthology.org/2023.findings-emnlp.986/

@inproceedings{bao-etal-2023-non, title = "Non-Autoregressive Document-Level Machine Translation", author = "Bao, Guangsheng and Teng, Zhiyang and Zhou, Hao and Yan, Jianhao and Zhang, Yue", 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.986", doi = "10.18653/v1/2023.findings-emnlp.986", pages = "14791--14803", abstract = "Non-autoregressive translation (NAT) models achieve comparable performance and superior speed compared to auto-regressive translation (AT) models in the context of sentence-level machine translation (MT). However, their abilities are unexplored in document-level MT, hindering their usage in real scenarios. In this paper, we conduct a comprehensive examination of typical NAT models in the context of document-level MT and further propose a simple but effective design of sentence alignment between source and target. Experiments show that NAT models achieve high acceleration on documents, and sentence alignment significantly enhances their performance. However, current NAT models still have a significant performance gap compared to their AT counterparts. Further investigation reveals that NAT models suffer more from the multi-modality and misalignment issues in the context of document-level MT, and current NAT models struggle with exploiting document context and handling discourse phenomena. We delve into these challenges and provide our code at \url{https://github.com/baoguangsheng/nat-on-doc}.", }

Non-autoregressive translation (NAT) models achieve comparable performance and superior speed compared to auto-regressive translation (AT) models in the context of sentence-level machine translation (MT). However, their abilities are unexplored in document-level MT, hindering their usage in real scenarios. In this paper, we conduct a comprehensive examination of typical NAT models in the context of document-level MT and further propose a simple but effective design of sentence alignment between source and target. Experiments show that NAT models achieve high acceleration on documents, and sentence alignment significantly enhances their performance. However, current NAT models still have a significant performance gap compared to their AT counterparts. Further investigation reveals that NAT models suffer more from the multi-modality and misalignment issues in the context of document-level MT, and current NAT models struggle with exploiting document context and handling discourse phenomena. We delve into these challenges and provide our code at \url{https://github.com/baoguangsheng/nat-on-doc}.

Non-Autoregressive Document-Level Machine Translation

findings-emnlp.986

2305.12878

Poster

https://aclanthology.org/2023.findings-emnlp.987.bib

https://aclanthology.org/2023.findings-emnlp.987/

@inproceedings{wu-etal-2023-exploring, title = "Exploring the Effectiveness of Multi-Lingual Commonsense Knowledge-Aware Open-Domain Dialogue Response Generation", author = "Wu, Sixing and Yu, Jiong and Che, Tianshi and Zhou, Yang and Zhou, Wei", 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.987", doi = "10.18653/v1/2023.findings-emnlp.987", pages = "14804--14814", abstract = "Prior works have shown the promising results of commonsense knowledge-aware models in improving informativeness while reducing the hallucination issue. Nonetheless, prior works often can only use monolingual knowledge whose language is consistent with the dialogue context. Except for a few high-resource languages, such as English and Chinese, most languages suffer from insufficient knowledge issues, especially minority languages. To this end, this work proposes a new task, Multi-Lingual Commonsense Knowledge-Aware Response Generation (MCKRG), which tries to use commonsense knowledge in other languages to enhance the current dialogue generation. Then, we construct a MCKRG dataset MCK-Dialog of seven languages with multiple alignment methods. Finally, we verify the effectiveness of using multi-lingual commonsense knowledge with a proposed MCK-T5 model. Extensive experimental results demonstrate the great potential of using multi-lingual commonsense knowledge in high-resource and low-resource languages. To the best of our knowledge, this work is the first to explore Multi-Lingual Commonsense Knowledge-Aware Response Generation.", }

Prior works have shown the promising results of commonsense knowledge-aware models in improving informativeness while reducing the hallucination issue. Nonetheless, prior works often can only use monolingual knowledge whose language is consistent with the dialogue context. Except for a few high-resource languages, such as English and Chinese, most languages suffer from insufficient knowledge issues, especially minority languages. To this end, this work proposes a new task, Multi-Lingual Commonsense Knowledge-Aware Response Generation (MCKRG), which tries to use commonsense knowledge in other languages to enhance the current dialogue generation. Then, we construct a MCKRG dataset MCK-Dialog of seven languages with multiple alignment methods. Finally, we verify the effectiveness of using multi-lingual commonsense knowledge with a proposed MCK-T5 model. Extensive experimental results demonstrate the great potential of using multi-lingual commonsense knowledge in high-resource and low-resource languages. To the best of our knowledge, this work is the first to explore Multi-Lingual Commonsense Knowledge-Aware Response Generation.

Exploring the Effectiveness of Multi-Lingual Commonsense Knowledge-Aware Open-Domain Dialogue Response Generation

findings-emnlp.987

Poster

https://aclanthology.org/2023.findings-emnlp.988.bib

https://aclanthology.org/2023.findings-emnlp.988/

@inproceedings{chen-etal-2023-mixture, title = "Mixture of Soft Prompts for Controllable Data Generation", author = "Chen, Derek and Lee, Celine and Lu, Yunan and Rosati, Domenic and Yu, Zhou", 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.988", doi = "10.18653/v1/2023.findings-emnlp.988", pages = "14815--14833", abstract = "Large language models (LLMs) effectively generate fluent text when the target output follows natural language patterns. However, structured prediction tasks confine the output format to a limited ontology, causing even very large models to struggle since they were never trained with such restrictions in mind. The difficulty of using LLMs for direct prediction is exacerbated in few-shot learning scenarios, which commonly arise due to domain shift and resource limitations. We flip the problem on its head by leveraging the LLM as a tool for data augmentation rather than direct prediction. Our proposed Mixture of Soft Prompts (MSP) serves as a parameter-efficient procedure for generating multi-attribute data in a controlled manner. Denoising mechanisms are further applied to improve the quality of synthesized data. Automatic metrics show our method is capable of producing diverse and natural text, while preserving label semantics. Moreover, MSP achieves state-of-the-art results on three benchmarks when compared against strong baselines. Our method offers an alternate data-centric approach for applying LLMs to complex prediction tasks.", }

Large language models (LLMs) effectively generate fluent text when the target output follows natural language patterns. However, structured prediction tasks confine the output format to a limited ontology, causing even very large models to struggle since they were never trained with such restrictions in mind. The difficulty of using LLMs for direct prediction is exacerbated in few-shot learning scenarios, which commonly arise due to domain shift and resource limitations. We flip the problem on its head by leveraging the LLM as a tool for data augmentation rather than direct prediction. Our proposed Mixture of Soft Prompts (MSP) serves as a parameter-efficient procedure for generating multi-attribute data in a controlled manner. Denoising mechanisms are further applied to improve the quality of synthesized data. Automatic metrics show our method is capable of producing diverse and natural text, while preserving label semantics. Moreover, MSP achieves state-of-the-art results on three benchmarks when compared against strong baselines. Our method offers an alternate data-centric approach for applying LLMs to complex prediction tasks.

Mixture of Soft Prompts for Controllable Data Generation

findings-emnlp.988

2303.01580

https://huggingface.co/papers/2303.01580

Poster

https://aclanthology.org/2023.findings-emnlp.989.bib

https://aclanthology.org/2023.findings-emnlp.989/

@inproceedings{tang-etal-2023-boundary, title = "A Boundary Offset Prediction Network for Named Entity Recognition", author = "Tang, Minghao and He, Yongquan and Xu, Yongxiu and Xu, Hongbo and Zhang, Wenyuan and Lin, Yang", 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.989", doi = "10.18653/v1/2023.findings-emnlp.989", pages = "14834--14846", abstract = "Named entity recognition (NER) is a fundamental task in natural language processing that aims to identify and classify named entities in text. However, span-based methods for NER typically assign entity types to text spans, resulting in an imbalanced sample space and neglecting the connections between non-entity and entity spans. To address these issues, we propose a novel approach for NER, named the Boundary Offset Prediction Network (BOPN), which predicts the boundary offsets between candidate spans and their nearest entity spans. By leveraging the guiding semantics of boundary offsets, BOPN establishes connections between non-entity and entity spans, enabling non-entity spans to function as additional positive samples for entity detection. Furthermore, our method integrates entity type and span representations to generate type-aware boundary offsets instead of using entity types as detection targets. We conduct experiments on eight widely-used NER datasets, and the results demonstrate that our proposed BOPN outperforms previous state-of-the-art methods.", }

Named entity recognition (NER) is a fundamental task in natural language processing that aims to identify and classify named entities in text. However, span-based methods for NER typically assign entity types to text spans, resulting in an imbalanced sample space and neglecting the connections between non-entity and entity spans. To address these issues, we propose a novel approach for NER, named the Boundary Offset Prediction Network (BOPN), which predicts the boundary offsets between candidate spans and their nearest entity spans. By leveraging the guiding semantics of boundary offsets, BOPN establishes connections between non-entity and entity spans, enabling non-entity spans to function as additional positive samples for entity detection. Furthermore, our method integrates entity type and span representations to generate type-aware boundary offsets instead of using entity types as detection targets. We conduct experiments on eight widely-used NER datasets, and the results demonstrate that our proposed BOPN outperforms previous state-of-the-art methods.

A Boundary Offset Prediction Network for Named Entity Recognition

findings-emnlp.989

2310.18349

Poster

https://aclanthology.org/2023.findings-emnlp.990.bib

https://aclanthology.org/2023.findings-emnlp.990/

@inproceedings{mai-etal-2023-prefix, title = "Prefix-Tuning Based Unsupervised Text Style Transfer", author = "Mai, Huiyu and Jiang, Wenhao and Deng, Zhi-Hong", 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.990", doi = "10.18653/v1/2023.findings-emnlp.990", pages = "14847--14856", abstract = "Unsupervised text style transfer aims at training a generative model that can alter the style of the input sentence while preserving its content without using any parallel data. In this paper, we employ powerful pre-trained large language models and present a new prefix-tuning-based method for unsupervised text style transfer. We construct three different kinds of prefixes, i.e., shared prefix, style prefix, and content prefix, to encode task-specific information, target style, and the content information of the input sentence, respectively. Compared to embeddings used by previous works, the proposed prefixes can provide richer information for the model. Furthermore, we adopt a recursive way of using language models in the process of style transfer. This strategy provides a more effective way for the interactions between the input sentence and GPT-2, helps the model construct more informative prefixes, and thus, helps improve the performance. Evaluations on the well-known datasets show that our method outperforms the state-of-the-art baselines. Results, analysis of ablation studies, and subjective evaluations from humans are also provided for a deeper understanding of the proposed method.", }

Unsupervised text style transfer aims at training a generative model that can alter the style of the input sentence while preserving its content without using any parallel data. In this paper, we employ powerful pre-trained large language models and present a new prefix-tuning-based method for unsupervised text style transfer. We construct three different kinds of prefixes, i.e., shared prefix, style prefix, and content prefix, to encode task-specific information, target style, and the content information of the input sentence, respectively. Compared to embeddings used by previous works, the proposed prefixes can provide richer information for the model. Furthermore, we adopt a recursive way of using language models in the process of style transfer. This strategy provides a more effective way for the interactions between the input sentence and GPT-2, helps the model construct more informative prefixes, and thus, helps improve the performance. Evaluations on the well-known datasets show that our method outperforms the state-of-the-art baselines. Results, analysis of ablation studies, and subjective evaluations from humans are also provided for a deeper understanding of the proposed method.

Prefix-Tuning Based Unsupervised Text Style Transfer

findings-emnlp.990

2310.14599

Poster

https://aclanthology.org/2023.findings-emnlp.991.bib

https://aclanthology.org/2023.findings-emnlp.991/

@inproceedings{chen-etal-2023-evaluating, title = "Evaluating and Enhancing the Robustness of Code Pre-trained Models through Structure-Aware Adversarial Samples Generation", author = "Chen, Nuo and Sun, Qiushi and Wang, Jianing and Gao, Ming and Li, Xiaoli and Li, Xiang", 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.991", doi = "10.18653/v1/2023.findings-emnlp.991", pages = "14857--14873", abstract = "Code pre-trained models (CodePTMs) have significantly advanced the field of neural code intelligence. Despite their capabilities, these models are susceptible to adversarial attacks that subtly modify the model inputs, resulting in incorrect outputs or predictions. Previous methods of robustness evaluation for CodePTMs primarily stem from a textual perspective, without explicitly taking into account the structure of the code. Furthermore, prior studies fail to encompass a broad enough spectrum of tasks and models. In this paper, we propose a set of novel robustness evaluation methods based on the intrinsic structure of the code. Specifically, we first launch adversarial attacks on crucial identifier tokens and sub-tree structures to explore the impact of imperceptible perturbation. Then, we perform global restructuring of the code using different traversal methods for abstract syntax trees, aiming to explore the model{'}s sensitivity to input samples with equivalent information. Moreover, for each scenario, we employ adversarial training methods to explore the possibility of restoring the performance of perturbed models. For both code understanding and generation, our proposed method has demonstrated its effectiveness across a wide range of models and tasks, thereby allowing us to make one step forward in our understanding of the inner mechanisms of CodePTMs.", }

Code pre-trained models (CodePTMs) have significantly advanced the field of neural code intelligence. Despite their capabilities, these models are susceptible to adversarial attacks that subtly modify the model inputs, resulting in incorrect outputs or predictions. Previous methods of robustness evaluation for CodePTMs primarily stem from a textual perspective, without explicitly taking into account the structure of the code. Furthermore, prior studies fail to encompass a broad enough spectrum of tasks and models. In this paper, we propose a set of novel robustness evaluation methods based on the intrinsic structure of the code. Specifically, we first launch adversarial attacks on crucial identifier tokens and sub-tree structures to explore the impact of imperceptible perturbation. Then, we perform global restructuring of the code using different traversal methods for abstract syntax trees, aiming to explore the model{'}s sensitivity to input samples with equivalent information. Moreover, for each scenario, we employ adversarial training methods to explore the possibility of restoring the performance of perturbed models. For both code understanding and generation, our proposed method has demonstrated its effectiveness across a wide range of models and tasks, thereby allowing us to make one step forward in our understanding of the inner mechanisms of CodePTMs.

Evaluating and Enhancing the Robustness of Code Pre-trained Models through Structure-Aware Adversarial Samples Generation

findings-emnlp.991

Poster

https://aclanthology.org/2023.findings-emnlp.992.bib

https://aclanthology.org/2023.findings-emnlp.992/

@inproceedings{kern-etal-2023-annotation, title = "Annotation Sensitivity: Training Data Collection Methods Affect Model Performance", author = "Kern, Christoph and Eckman, Stephanie and Beck, Jacob and Chew, Rob and Ma, Bolei and Kreuter, Frauke", 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.992", doi = "10.18653/v1/2023.findings-emnlp.992", pages = "14874--14886", abstract = "When training data are collected from human annotators, the design of the annotation instrument, the instructions given to annotators, the characteristics of the annotators, and their interactions can impact training data. This study demonstrates that design choices made when creating an annotation instrument also impact the models trained on the resulting annotations. We introduce the term annotation sensitivity to refer to the impact of annotation data collection methods on the annotations themselves and on downstream model performance and predictions. We collect annotations of hate speech and offensive language in five experimental conditions of an annotation instrument, randomly assigning annotators to conditions. We then fine-tune BERT models on each of the five resulting datasets and evaluate model performance on a holdout portion of each condition. We find considerable differences between the conditions for 1) the share of hate speech/offensive language annotations, 2) model performance, 3) model predictions, and 4) model learning curves. Our results emphasize the crucial role played by the annotation instrument which has received little attention in the machine learning literature. We call for additional research into how and why the instrument impacts the annotations to inform the development of best practices in instrument design.", }

When training data are collected from human annotators, the design of the annotation instrument, the instructions given to annotators, the characteristics of the annotators, and their interactions can impact training data. This study demonstrates that design choices made when creating an annotation instrument also impact the models trained on the resulting annotations. We introduce the term annotation sensitivity to refer to the impact of annotation data collection methods on the annotations themselves and on downstream model performance and predictions. We collect annotations of hate speech and offensive language in five experimental conditions of an annotation instrument, randomly assigning annotators to conditions. We then fine-tune BERT models on each of the five resulting datasets and evaluate model performance on a holdout portion of each condition. We find considerable differences between the conditions for 1) the share of hate speech/offensive language annotations, 2) model performance, 3) model predictions, and 4) model learning curves. Our results emphasize the crucial role played by the annotation instrument which has received little attention in the machine learning literature. We call for additional research into how and why the instrument impacts the annotations to inform the development of best practices in instrument design.

Annotation Sensitivity: Training Data Collection Methods Affect Model Performance

findings-emnlp.992

2311.14212

Poster

https://aclanthology.org/2023.findings-emnlp.993.bib

https://aclanthology.org/2023.findings-emnlp.993/

@inproceedings{piano-etal-2023-qualitative, title = "Qualitative Code Suggestion: A Human-Centric Approach to Qualitative Coding", author = "Spinoso-Di Piano, Cesare and Rahimi, Samira and Cheung, Jackie", 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.993", doi = "10.18653/v1/2023.findings-emnlp.993", pages = "14887--14909", abstract = "Qualitative coding is a content analysis method in which researchers read through a text corpus and assign descriptive labels or qualitative codes to passages. It is an arduous and manual process which human-computer interaction (HCI) studies have shown could greatly benefit from NLP techniques to assist qualitative coders. Yet, previous attempts at leveraging language technologies have set up qualitative coding as a fully automatable classification problem. In this work, we take a more assistive approach by defining the task of qualitative code suggestion (QCS) in which a ranked list of previously assigned qualitative codes is suggested from an identified passage. In addition to being user-motivated, QCS integrates previously ignored properties of qualitative coding such as the sequence in which passages are annotated, the importance of rare codes and the differences in annotation styles between coders. We investigate the QCS task by releasing the first publicly available qualitative coding dataset, CVDQuoding, consisting of interviews conducted with women at risk of cardiovascular disease. In addition, we conduct a human evaluation which shows that our systems consistently make relevant code suggestions.", }

Qualitative coding is a content analysis method in which researchers read through a text corpus and assign descriptive labels or qualitative codes to passages. It is an arduous and manual process which human-computer interaction (HCI) studies have shown could greatly benefit from NLP techniques to assist qualitative coders. Yet, previous attempts at leveraging language technologies have set up qualitative coding as a fully automatable classification problem. In this work, we take a more assistive approach by defining the task of qualitative code suggestion (QCS) in which a ranked list of previously assigned qualitative codes is suggested from an identified passage. In addition to being user-motivated, QCS integrates previously ignored properties of qualitative coding such as the sequence in which passages are annotated, the importance of rare codes and the differences in annotation styles between coders. We investigate the QCS task by releasing the first publicly available qualitative coding dataset, CVDQuoding, consisting of interviews conducted with women at risk of cardiovascular disease. In addition, we conduct a human evaluation which shows that our systems consistently make relevant code suggestions.

Qualitative Code Suggestion: A Human-Centric Approach to Qualitative Coding

findings-emnlp.993

Poster

https://aclanthology.org/2023.findings-emnlp.994.bib

https://aclanthology.org/2023.findings-emnlp.994/

@inproceedings{liang-etal-2023-d2tv, title = "{D}$^2${TV}: Dual Knowledge Distillation and Target-oriented Vision Modeling for Many-to-Many Multimodal Summarization", author = "Liang, Yunlong and Meng, Fandong and Wang, Jiaan and Xu, Jinan and Chen, Yufeng and Zhou, 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.994", doi = "10.18653/v1/2023.findings-emnlp.994", pages = "14910--14922", abstract = "Many-to-many multimodal summarization (M$^3$S) task aims to generate summaries in any language with document inputs in any language and the corresponding image sequence, which essentially comprises of multimodal monolingual summarization (MMS) and multimodal cross-lingual summarization (MXLS) tasks. Although much work has been devoted to either MMS or MXLS, little research pays attention to the M$^3$S task. Besides, existing studies mainly focus on 1) utilizing MMS to enhance MXLS via knowledge distillation without considering the performance of MMS or 2) improving MMS models by filtering summary-unrelated visual features with implicit learning or explicitly complex training objectives. In this paper, we first introduce a general and practical task, \textit{i.e.}, M$^3$S. Further, we propose a dual knowledge distillation and target-oriented vision modeling framework for the M$^3$S task. Specifically, the dual knowledge distillation method guarantees that the knowledge of MMS and MXLS can be transferred to each other and thus mutually prompt both of them. To offer target-oriented visual features, a simple yet effective target-oriented contrastive objective is designed and responsible for discarding needless visual information. Extensive experiments on the many-to-many setting show the effectiveness of the proposed approach. Additionally, we contribute a many-to-many multimodal summarization (lmttM$^3$Sum) dataset with 44 languages to facilitate future research.", }

Many-to-many multimodal summarization (M$^3$S) task aims to generate summaries in any language with document inputs in any language and the corresponding image sequence, which essentially comprises of multimodal monolingual summarization (MMS) and multimodal cross-lingual summarization (MXLS) tasks. Although much work has been devoted to either MMS or MXLS, little research pays attention to the M$^3$S task. Besides, existing studies mainly focus on 1) utilizing MMS to enhance MXLS via knowledge distillation without considering the performance of MMS or 2) improving MMS models by filtering summary-unrelated visual features with implicit learning or explicitly complex training objectives. In this paper, we first introduce a general and practical task, \textit{i.e.}, M$^3$S. Further, we propose a dual knowledge distillation and target-oriented vision modeling framework for the M$^3$S task. Specifically, the dual knowledge distillation method guarantees that the knowledge of MMS and MXLS can be transferred to each other and thus mutually prompt both of them. To offer target-oriented visual features, a simple yet effective target-oriented contrastive objective is designed and responsible for discarding needless visual information. Extensive experiments on the many-to-many setting show the effectiveness of the proposed approach. Additionally, we contribute a many-to-many multimodal summarization (lmttM$^3$Sum) dataset with 44 languages to facilitate future research.

D^2TV: Dual Knowledge Distillation and Target-oriented Vision Modeling for Many-to-Many Multimodal Summarization

findings-emnlp.994

Poster

https://aclanthology.org/2023.findings-emnlp.995.bib

https://aclanthology.org/2023.findings-emnlp.995/

@inproceedings{gong-etal-2023-improving, title = "Improving Input-label Mapping with Demonstration Replay for In-context Learning", author = "Gong, Zhuocheng and Liu, Jiahao and Wang, Qifan and Wang, Jingang and Cai, Xunliang and Zhao, Dongyan and Yan, Rui", 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.995", doi = "10.18653/v1/2023.findings-emnlp.995", pages = "14923--14934", abstract = "In-context learning (ICL) is an emerging capability of large autoregressive language models where a few input-label demonstrations are appended to the input to enhance the model{'}s understanding of downstream NLP tasks, without directly adjusting the model parameters. The effectiveness of ICL can be attributed to the strong language modeling capabilities of large language models (LLMs), which enable them to learn the mapping between input and labels based on in-context demonstrations. Despite achieving promising results, the causal nature of language modeling in ICL restricts the attention to be backward only, i.e., a token only attends to its previous tokens, failing to capture the full input-label information and limiting the model{'}s performance. In this paper, we propose a novel ICL method called Repeated Demonstration with Sliding Causal Attention, (RdSca). Specifically, we duplicate later demonstrations and concatenate them to the front, allowing the model to {`}observe{'} the later information even under the causal restriction. Besides, we introduce sliding causal attention, which customizes causal attention to avoid information leakage. Experimental results show that our method significantly improves the input-label mapping in ICL demonstrations. We also conduct an in-depth analysis of how to customize the causal attention without training, which has been an unexplored area in previous research.", }

In-context learning (ICL) is an emerging capability of large autoregressive language models where a few input-label demonstrations are appended to the input to enhance the model{'}s understanding of downstream NLP tasks, without directly adjusting the model parameters. The effectiveness of ICL can be attributed to the strong language modeling capabilities of large language models (LLMs), which enable them to learn the mapping between input and labels based on in-context demonstrations. Despite achieving promising results, the causal nature of language modeling in ICL restricts the attention to be backward only, i.e., a token only attends to its previous tokens, failing to capture the full input-label information and limiting the model{'}s performance. In this paper, we propose a novel ICL method called Repeated Demonstration with Sliding Causal Attention, (RdSca). Specifically, we duplicate later demonstrations and concatenate them to the front, allowing the model to {`}observe{'} the later information even under the causal restriction. Besides, we introduce sliding causal attention, which customizes causal attention to avoid information leakage. Experimental results show that our method significantly improves the input-label mapping in ICL demonstrations. We also conduct an in-depth analysis of how to customize the causal attention without training, which has been an unexplored area in previous research.

Improving Input-label Mapping with Demonstration Replay for In-context Learning

findings-emnlp.995

2310.19572

Poster

https://aclanthology.org/2023.findings-emnlp.996.bib

https://aclanthology.org/2023.findings-emnlp.996/

@inproceedings{nan-etal-2023-enhancing, title = "Enhancing Text-to-{SQL} Capabilities of Large Language Models: A Study on Prompt Design Strategies", author = "Nan, Linyong and Zhao, Yilun and Zou, Weijin and Ri, Narutatsu and Tae, Jaesung and Zhang, Ellen and Cohan, Arman and Radev, Dragomir", 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.996", doi = "10.18653/v1/2023.findings-emnlp.996", pages = "14935--14956", abstract = "In-context learning (ICL) has emerged as a new approach to various natural language processing tasks, utilizing large language models (LLMs) to make predictions based on context that has been supplemented with a few examples or task-specific instructions. In this paper, we aim to extend this method to question answering tasks that utilize structured knowledge sources, and improve Text-to-SQL systems by exploring various prompt design strategies for employing LLMs. We conduct a systematic investigation into different demonstration selection methods and optimal instruction formats for prompting LLMs in the Text-to-SQL task. Our approach involves leveraging the syntactic structure of an example{'}s SQL query to retrieve demonstrations, and we demonstrate that pursuing both diversity and similarity in demonstration selection leads to enhanced performance. Furthermore, we show that LLMs benefit from database-related knowledge augmentations. Our most effective strategy outperforms the state-of-the-art system by 2.5 points (Execution Accuracy) and the best fine-tuned system by 5.1 points on the Spider dataset. These results highlight the effectiveness of our approach in adapting LLMs to the Text-to-SQL task, and we present an analysis of the factors contributing to the success of our strategy.", }

In-context learning (ICL) has emerged as a new approach to various natural language processing tasks, utilizing large language models (LLMs) to make predictions based on context that has been supplemented with a few examples or task-specific instructions. In this paper, we aim to extend this method to question answering tasks that utilize structured knowledge sources, and improve Text-to-SQL systems by exploring various prompt design strategies for employing LLMs. We conduct a systematic investigation into different demonstration selection methods and optimal instruction formats for prompting LLMs in the Text-to-SQL task. Our approach involves leveraging the syntactic structure of an example{'}s SQL query to retrieve demonstrations, and we demonstrate that pursuing both diversity and similarity in demonstration selection leads to enhanced performance. Furthermore, we show that LLMs benefit from database-related knowledge augmentations. Our most effective strategy outperforms the state-of-the-art system by 2.5 points (Execution Accuracy) and the best fine-tuned system by 5.1 points on the Spider dataset. These results highlight the effectiveness of our approach in adapting LLMs to the Text-to-SQL task, and we present an analysis of the factors contributing to the success of our strategy.

Enhancing Text-to-SQL Capabilities of Large Language Models: A Study on Prompt Design Strategies

findings-emnlp.996

https://huggingface.co/papers/2305.12586

Poster

https://aclanthology.org/2023.findings-emnlp.997.bib

https://aclanthology.org/2023.findings-emnlp.997/

@inproceedings{ogundepo-etal-2023-cross, title = "Cross-lingual Open-Retrieval Question Answering for {A}frican Languages", author = "Ogundepo, Odunayo and Gwadabe, Tajuddeen and Rivera, Clara and Clark, Jonathan and Ruder, Sebastian and Adelani, David and Dossou, Bonaventure and Diop, Abdou and Sikasote, Claytone and Hacheme, Gilles and Buzaaba, Happy and Ezeani, Ignatius and Mabuya, Rooweither and Osei, Salomey and Emezue, Chris and Kahira, Albert and Muhammad, Shamsuddeen and Oladipo, Akintunde and Owodunni, Abraham and Tonja, Atnafu and Shode, Iyanuoluwa and Asai, Akari and Aremu, Anuoluwapo and Awokoya, Ayodele and Opoku, Bernard and Chukwuneke, Chiamaka and Mwase, Christine and Siro, Clemencia and Arthur, Stephen and Ajayi, Tunde and Otiende, Verrah and Rubungo, Andre and Sinkala, Boyd and Ajisafe, Daniel and Onwuegbuzia, Emeka and Lawan, Falalu and Ahmad, Ibrahim and Alabi, Jesujoba and Mbonu, Chinedu and Adeyemi, Mofetoluwa and Phiri, Mofya and Ahia, Orevaoghene and Iro, Ruqayya and Adhiambo, Sonia", 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.997", doi = "10.18653/v1/2023.findings-emnlp.997", pages = "14957--14972", abstract = "African languages have far less in-language content available digitally, making it challenging for question answering systems to satisfy the information needs of users. Cross-lingual open-retrieval question answering (XOR QA) systems {--} those that retrieve answer content from other languages while serving people in their native language{---}offer a means of filling this gap. To this end, we create Our Dataset, the first cross-lingual QA dataset with a focus on African languages. Our Dataset includes 12,000+ XOR QA examples across 10 African languages. While previous datasets have focused primarily on languages where cross-lingual QA augments coverage from the target language, Our Dataset focuses on languages where cross-lingual answer content is the only high-coverage source of answer content. Because of this, we argue that African languages are one of the most important and realistic use cases for XOR QA. Our experiments demonstrate the poor performance of automatic translation and multilingual retrieval methods. Overall, Our Dataset proves challenging for state-of-the-art QA models. We hope that the dataset enables the development of more equitable QA technology.", }

African languages have far less in-language content available digitally, making it challenging for question answering systems to satisfy the information needs of users. Cross-lingual open-retrieval question answering (XOR QA) systems {--} those that retrieve answer content from other languages while serving people in their native language{---}offer a means of filling this gap. To this end, we create Our Dataset, the first cross-lingual QA dataset with a focus on African languages. Our Dataset includes 12,000+ XOR QA examples across 10 African languages. While previous datasets have focused primarily on languages where cross-lingual QA augments coverage from the target language, Our Dataset focuses on languages where cross-lingual answer content is the only high-coverage source of answer content. Because of this, we argue that African languages are one of the most important and realistic use cases for XOR QA. Our experiments demonstrate the poor performance of automatic translation and multilingual retrieval methods. Overall, Our Dataset proves challenging for state-of-the-art QA models. We hope that the dataset enables the development of more equitable QA technology.

Cross-lingual Open-Retrieval Question Answering for African Languages

findings-emnlp.997

https://huggingface.co/papers/2305.06897

Poster

https://aclanthology.org/2023.findings-emnlp.998.bib

https://aclanthology.org/2023.findings-emnlp.998/

@inproceedings{stap-etal-2023-viewing, title = "Viewing Knowledge Transfer in Multilingual Machine Translation Through a Representational Lens", author = "Stap, David and Niculae, Vlad and Monz, Christof", 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.998", doi = "10.18653/v1/2023.findings-emnlp.998", pages = "14973--14987", abstract = "We argue that translation quality alone is not a sufficient metric for measuring knowledge transfer in multilingual neural machine translation. To support this claim, we introduce Representational Transfer Potential (RTP), which measures representational similarities between languages. We show that RTP can measure both positive and negative transfer (interference), and find that RTP is strongly correlated with changes in translation quality, indicating that transfer \textit{does} occur. Furthermore, we investigate data and language characteristics that are relevant for transfer, and find that multi-parallel overlap is an important yet under-explored feature. Based on this, we develop a novel training scheme, which uses an auxiliary similarity loss that encourages representations to be more invariant across languages by taking advantage of multi-parallel data. We show that our method yields increased translation quality for low- and mid-resource languages across multiple data and model setups.", }

We argue that translation quality alone is not a sufficient metric for measuring knowledge transfer in multilingual neural machine translation. To support this claim, we introduce Representational Transfer Potential (RTP), which measures representational similarities between languages. We show that RTP can measure both positive and negative transfer (interference), and find that RTP is strongly correlated with changes in translation quality, indicating that transfer \textit{does} occur. Furthermore, we investigate data and language characteristics that are relevant for transfer, and find that multi-parallel overlap is an important yet under-explored feature. Based on this, we develop a novel training scheme, which uses an auxiliary similarity loss that encourages representations to be more invariant across languages by taking advantage of multi-parallel data. We show that our method yields increased translation quality for low- and mid-resource languages across multiple data and model setups.

Viewing Knowledge Transfer in Multilingual Machine Translation Through a Representational Lens

findings-emnlp.998

2305.11550

Poster

https://aclanthology.org/2023.findings-emnlp.999.bib

https://aclanthology.org/2023.findings-emnlp.999/

@inproceedings{naszadi-etal-2023-aligning, title = "Aligning Predictive Uncertainty with Clarification Questions in Grounded Dialog", author = "Naszadi, Kata and Manggala, Putra and Monz, Christof", 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.999", doi = "10.18653/v1/2023.findings-emnlp.999", pages = "14988--14998", abstract = "Asking for clarification is fundamental to effective collaboration. An interactive artificial agent must know when to ask a human instructor for more information in order to ascertain their goals. Previous work bases the timing of questions on supervised models learned from interactions between humans. Instead of a supervised classification task, we wish to ground the need for questions in the acting agent{'}s predictive uncertainty. In this work, we investigate if ambiguous linguistic instructions can be aligned with uncertainty in neural models. We train an agent using the T5 encoder-decoder architecture to solve the Minecraft Collaborative Building Task and identify uncertainty metrics that achieve better distributional separation between clear and ambiguous instructions. We further show that well-calibrated prediction probabilities benefit the detection of ambiguous instructions. Lastly, we provide a novel empirical analysis on the relationship between uncertainty and dialog history length and highlight an important property that poses a difficulty for detection.", }

Asking for clarification is fundamental to effective collaboration. An interactive artificial agent must know when to ask a human instructor for more information in order to ascertain their goals. Previous work bases the timing of questions on supervised models learned from interactions between humans. Instead of a supervised classification task, we wish to ground the need for questions in the acting agent{'}s predictive uncertainty. In this work, we investigate if ambiguous linguistic instructions can be aligned with uncertainty in neural models. We train an agent using the T5 encoder-decoder architecture to solve the Minecraft Collaborative Building Task and identify uncertainty metrics that achieve better distributional separation between clear and ambiguous instructions. We further show that well-calibrated prediction probabilities benefit the detection of ambiguous instructions. Lastly, we provide a novel empirical analysis on the relationship between uncertainty and dialog history length and highlight an important property that poses a difficulty for detection.

Aligning Predictive Uncertainty with Clarification Questions in Grounded Dialog

findings-emnlp.999

Poster

https://aclanthology.org/2023.findings-emnlp.1000.bib

https://aclanthology.org/2023.findings-emnlp.1000/

@inproceedings{stogiannidis-etal-2023-cache, title = "Cache me if you Can: an Online Cost-aware Teacher-Student framework to Reduce the Calls to Large Language Models", author = "Stogiannidis, Ilias and Vassos, Stavros and Malakasiotis, Prodromos and Androutsopoulos, Ion", 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.1000", doi = "10.18653/v1/2023.findings-emnlp.1000", pages = "14999--15008", abstract = "Prompting Large Language Models (LLMs) performs impressively in zero- and few-shot settings. Hence, small and medium-sized enterprises (SMEs) that cannot afford the cost of creating large task-specific training datasets, but also the cost of pretraining their own LLMs, are increasingly turning to third-party services that allow them to prompt LLMs. However, such services currently require a payment per call, which becomes a significant operating expense (OpEx). Furthermore, customer inputs are often very similar over time, hence SMEs end-up prompting LLMs with very similar instances. We propose a framework that allows reducing the calls to LLMs by caching previous LLM responses and using them to train a local inexpensive model on the SME side. The framework includes criteria for deciding when to trust the local model or call the LLM, and a methodology to tune the criteria and measure the tradeoff between performance and cost. For experimental purposes, we instantiate our framework with two LLMs, GPT-3.5 or GPT-4, and two inexpensive students, a $k$-NN classifier or a Multi-Layer Perceptron, using two common business tasks, intent recognition and sentiment analysis. Experimental results indicate that significant OpEx savings can be obtained with only slightly lower performance.", }

Prompting Large Language Models (LLMs) performs impressively in zero- and few-shot settings. Hence, small and medium-sized enterprises (SMEs) that cannot afford the cost of creating large task-specific training datasets, but also the cost of pretraining their own LLMs, are increasingly turning to third-party services that allow them to prompt LLMs. However, such services currently require a payment per call, which becomes a significant operating expense (OpEx). Furthermore, customer inputs are often very similar over time, hence SMEs end-up prompting LLMs with very similar instances. We propose a framework that allows reducing the calls to LLMs by caching previous LLM responses and using them to train a local inexpensive model on the SME side. The framework includes criteria for deciding when to trust the local model or call the LLM, and a methodology to tune the criteria and measure the tradeoff between performance and cost. For experimental purposes, we instantiate our framework with two LLMs, GPT-3.5 or GPT-4, and two inexpensive students, a $k$-NN classifier or a Multi-Layer Perceptron, using two common business tasks, intent recognition and sentiment analysis. Experimental results indicate that significant OpEx savings can be obtained with only slightly lower performance.

Cache me if you Can: an Online Cost-aware Teacher-Student framework to Reduce the Calls to Large Language Models

findings-emnlp.1000

2310.13395

Poster

https://aclanthology.org/2023.findings-emnlp.1001.bib

https://aclanthology.org/2023.findings-emnlp.1001/

@inproceedings{jiao-etal-2023-parrot, title = "{P}arro{T}: Translating during Chat using Large Language Models tuned with Human Translation and Feedback", author = "Jiao, Wenxiang and Huang, Jen-tse and Wang, Wenxuan and He, Zhiwei and Liang, Tian and Wang, Xing and Shi, Shuming and Tu, Zhaopeng", 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.1001", doi = "10.18653/v1/2023.findings-emnlp.1001", pages = "15009--15020", abstract = "Large language models (LLMs) like ChatGPT have exhibited remarkable abilities on a wide range of natural language processing (NLP) tasks, including various machine translation abilities accomplished during chat. However, these models are only accessible through restricted APIs, which creates barriers to new research and advancements in the field. Therefore, we propose ParroT, a framework to enhance and regulate the translation abilities during chat based on open-source LLMs (e.g., LLaMA), human-written translation and feedback data. Specifically, ParroT reformulates translation data into the instruction-following style, and introduces a {``}Hint{''} field for incorporating extra requirements to regulate the translation process. Accordingly, we propose three instruction types for finetuning ParroT models, including translation instruction, contrastive instruction, and error-guided instruction. Experiments on Flores subsets and WMT22 test sets suggest that translation instruction improves the translation performance of vanilla LLMs significantly while error-guided instruction can lead to further improvement, which demonstrates the importance of learning from low-quality translations annotated by humans. We also demonstrate the potential of automatic evaluation tools in providing quality information of translations, when constructing error-guided instructions for directions that lack human annotation data. Please refer to our Github project for more implementation details: https://github.com/wxjiao/ParroT.", }

Large language models (LLMs) like ChatGPT have exhibited remarkable abilities on a wide range of natural language processing (NLP) tasks, including various machine translation abilities accomplished during chat. However, these models are only accessible through restricted APIs, which creates barriers to new research and advancements in the field. Therefore, we propose ParroT, a framework to enhance and regulate the translation abilities during chat based on open-source LLMs (e.g., LLaMA), human-written translation and feedback data. Specifically, ParroT reformulates translation data into the instruction-following style, and introduces a {``}Hint{''} field for incorporating extra requirements to regulate the translation process. Accordingly, we propose three instruction types for finetuning ParroT models, including translation instruction, contrastive instruction, and error-guided instruction. Experiments on Flores subsets and WMT22 test sets suggest that translation instruction improves the translation performance of vanilla LLMs significantly while error-guided instruction can lead to further improvement, which demonstrates the importance of learning from low-quality translations annotated by humans. We also demonstrate the potential of automatic evaluation tools in providing quality information of translations, when constructing error-guided instructions for directions that lack human annotation data. Please refer to our Github project for more implementation details: https://github.com/wxjiao/ParroT.

ParroT: Translating during Chat using Large Language Models tuned with Human Translation and Feedback

findings-emnlp.1001

2304.02426

Poster

https://aclanthology.org/2023.findings-emnlp.1002.bib

https://aclanthology.org/2023.findings-emnlp.1002/

@inproceedings{xu-etal-2023-dense, title = "Dense Retrieval as Indirect Supervision for Large-space Decision Making", author = "Xu, Nan and Wang, Fei and Dong, Mingtao and Chen, Muhao", 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.1002", doi = "10.18653/v1/2023.findings-emnlp.1002", pages = "15021--15033", abstract = "Many discriminative natural language understanding (NLU) tasks have large label spaces. Learning such a process of large-space decision making is particularly challenging due to the lack of training instances per label and the difficulty of selection among many fine-grained labels. Inspired by dense retrieval methods for passage finding in open-domain QA, we propose a reformulation of large-space discriminative NLU tasks as a learning-to-retrieve task, leading to a novel solution named Dense Decision Retrieval (DDR). Instead of predicting fine-grained decisions as logits, DDR adopts a dual-encoder architecture that learns to predict by retrieving from a decision thesaurus. This approach not only leverages rich indirect supervision signals from easy-to-consume learning resources for dense retrieval, it also leads to enhanced prediction generalizability with a semantically meaningful representation of the large decision space. When evaluated on tasks with decision spaces ranging from hundreds to hundred-thousand scales, DDR outperforms strong baselines greatly by 27.54{\%} in P @1 on two extreme multi-label classification tasks, 1.17{\%} in F1 score ultra-fine entity typing, and 1.26{\%} in accuracy on three few-shot intent classification tasks on average.", }

Many discriminative natural language understanding (NLU) tasks have large label spaces. Learning such a process of large-space decision making is particularly challenging due to the lack of training instances per label and the difficulty of selection among many fine-grained labels. Inspired by dense retrieval methods for passage finding in open-domain QA, we propose a reformulation of large-space discriminative NLU tasks as a learning-to-retrieve task, leading to a novel solution named Dense Decision Retrieval (DDR). Instead of predicting fine-grained decisions as logits, DDR adopts a dual-encoder architecture that learns to predict by retrieving from a decision thesaurus. This approach not only leverages rich indirect supervision signals from easy-to-consume learning resources for dense retrieval, it also leads to enhanced prediction generalizability with a semantically meaningful representation of the large decision space. When evaluated on tasks with decision spaces ranging from hundreds to hundred-thousand scales, DDR outperforms strong baselines greatly by 27.54{\%} in P @1 on two extreme multi-label classification tasks, 1.17{\%} in F1 score ultra-fine entity typing, and 1.26{\%} in accuracy on three few-shot intent classification tasks on average.

Dense Retrieval as Indirect Supervision for Large-space Decision Making

findings-emnlp.1002

2310.18619

Poster

https://aclanthology.org/2023.findings-emnlp.1003.bib

https://aclanthology.org/2023.findings-emnlp.1003/

@inproceedings{ma-etal-2023-one, title = "One-Model-Connects-All: A Unified Graph Pre-Training Model for Online Community Modeling", author = "Ma, Ruoxue and Xu, Jiarong and Zhang, Xinnong and Zhang, Haozhe and Zhao, Zuyu and Zhang, Qi and Huang, Xuanjing and Wei, Zhongyu", 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.1003", doi = "10.18653/v1/2023.findings-emnlp.1003", pages = "15034--15045", abstract = "Online community is composed of communities, users, and user-generated textual content, with rich information that can help us solve social problems. Previous research hasn{'}t fully utilized these three components and the relationship among them. What{'}s more, they can{'}t adapt to a wide range of downstream tasks. To solve these problems, we focus on a framework that simultaneously considers communities, users, and texts. And it can easily connect with a variety of downstream tasks related to social media. Specifically, we use a ternary heterogeneous graph to model online communities. Text reconstruction and edge generation are used to learn structural and semantic knowledge among communities, users, and texts. By leveraging this pre-trained model, we achieve promising results across multiple downstream tasks, such as violation detection, sentiment analysis, and community recommendation. Our exploration will improve online community modeling.", }

Online community is composed of communities, users, and user-generated textual content, with rich information that can help us solve social problems. Previous research hasn{'}t fully utilized these three components and the relationship among them. What{'}s more, they can{'}t adapt to a wide range of downstream tasks. To solve these problems, we focus on a framework that simultaneously considers communities, users, and texts. And it can easily connect with a variety of downstream tasks related to social media. Specifically, we use a ternary heterogeneous graph to model online communities. Text reconstruction and edge generation are used to learn structural and semantic knowledge among communities, users, and texts. By leveraging this pre-trained model, we achieve promising results across multiple downstream tasks, such as violation detection, sentiment analysis, and community recommendation. Our exploration will improve online community modeling.

One-Model-Connects-All: A Unified Graph Pre-Training Model for Online Community Modeling

findings-emnlp.1003

Poster

https://aclanthology.org/2023.findings-emnlp.1004.bib

https://aclanthology.org/2023.findings-emnlp.1004/

@inproceedings{tian-etal-2023-image, title = "In-Image Neural Machine Translation with Segmented Pixel Sequence-to-Sequence Model", author = "Tian, Yanzhi and Li, Xiang and Liu, Zeming and Guo, Yuhang and Wang, Bin", 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.1004", doi = "10.18653/v1/2023.findings-emnlp.1004", pages = "15046--15057", abstract = "In-Image Machine Translation (IIMT) aims to convert images containing texts from one language to another. Traditional approaches for this task are cascade methods, which utilize optical character recognition (OCR) followed by neural machine translation (NMT) and text rendering. However, the cascade methods suffer from compounding errors of OCR and NMT, leading to a decrease in translation quality. In this paper, we propose an end-to-end model instead of the OCR, NMT and text rendering pipeline. Our neural architecture adopts encoder-decoder paradigm with segmented pixel sequences as inputs and outputs. Through end-to-end training, our model yields improvements across various dimensions, (i) it achieves higher translation quality by avoiding error propagation, (ii) it demonstrates robustness for out domain data, and (iii) it displays insensitivity to incomplete words. To validate the effectiveness of our method and support for future research, we construct our dataset containing 4M pairs of De-En images and train our end-to-end model. The experimental results show that our approach outperforms both cascade method and current end-to-end model.", }

In-Image Machine Translation (IIMT) aims to convert images containing texts from one language to another. Traditional approaches for this task are cascade methods, which utilize optical character recognition (OCR) followed by neural machine translation (NMT) and text rendering. However, the cascade methods suffer from compounding errors of OCR and NMT, leading to a decrease in translation quality. In this paper, we propose an end-to-end model instead of the OCR, NMT and text rendering pipeline. Our neural architecture adopts encoder-decoder paradigm with segmented pixel sequences as inputs and outputs. Through end-to-end training, our model yields improvements across various dimensions, (i) it achieves higher translation quality by avoiding error propagation, (ii) it demonstrates robustness for out domain data, and (iii) it displays insensitivity to incomplete words. To validate the effectiveness of our method and support for future research, we construct our dataset containing 4M pairs of De-En images and train our end-to-end model. The experimental results show that our approach outperforms both cascade method and current end-to-end model.

In-Image Neural Machine Translation with Segmented Pixel Sequence-to-Sequence Model

findings-emnlp.1004

Poster

https://aclanthology.org/2023.findings-emnlp.1005.bib

https://aclanthology.org/2023.findings-emnlp.1005/

@inproceedings{moskvichev-mai-2023-narrativexl, title = "{N}arrative{XL}: a Large-scale Dataset for Long-Term Memory Models", author = "Moskvichev, Arsenii and Mai, Ky-Vinh", 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.1005", doi = "10.18653/v1/2023.findings-emnlp.1005", pages = "15058--15072", abstract = "We propose a new large-scale (nearly a million questions) ultra-long-context (more than 50,000 words average document length) reading comprehension dataset. Using GPT 3.5, we summarized each scene in 1,500 hand-curated fiction books from Project Gutenberg, which resulted in approximately 150 scene-level summaries per book. After that, we created a number of reading comprehension questions based on these summaries, including three types of multiple-choice scene recognition questions, as well as free-form narrative reconstruction questions. With 990,595 total questions, our dataset is an order of magnitude larger than the closest alternatives. Crucially, most questions have a known {``}retention demand{''}, indicating how long-term of a memory is needed to answer them, which should aid long-term memory performance evaluation. We validate our data in four small-scale experiments: one with human labelers, and three with existing language models. We show that our questions 1) adequately represent the source material 2) can be used to diagnose a model{'}s memory capacity 3) are not trivial for modern language models even when the memory demand does not exceed those models{'} context lengths. Lastly, we provide our code which can be used to further expand the dataset with minimal human labor.", }

We propose a new large-scale (nearly a million questions) ultra-long-context (more than 50,000 words average document length) reading comprehension dataset. Using GPT 3.5, we summarized each scene in 1,500 hand-curated fiction books from Project Gutenberg, which resulted in approximately 150 scene-level summaries per book. After that, we created a number of reading comprehension questions based on these summaries, including three types of multiple-choice scene recognition questions, as well as free-form narrative reconstruction questions. With 990,595 total questions, our dataset is an order of magnitude larger than the closest alternatives. Crucially, most questions have a known {``}retention demand{''}, indicating how long-term of a memory is needed to answer them, which should aid long-term memory performance evaluation. We validate our data in four small-scale experiments: one with human labelers, and three with existing language models. We show that our questions 1) adequately represent the source material 2) can be used to diagnose a model{'}s memory capacity 3) are not trivial for modern language models even when the memory demand does not exceed those models{'} context lengths. Lastly, we provide our code which can be used to further expand the dataset with minimal human labor.

NarrativeXL: a Large-scale Dataset for Long-Term Memory Models

findings-emnlp.1005

2305.13877

Poster

https://aclanthology.org/2023.findings-emnlp.1006.bib

https://aclanthology.org/2023.findings-emnlp.1006/

@inproceedings{liermann-etal-2023-dialogue, title = "Dialogue Act-Aided Backchannel Prediction Using Multi-Task Learning", author = "Liermann, Wencke and Park, Yo-Han and Choi, Yong-Seok and Lee, Kong", 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.1006", doi = "10.18653/v1/2023.findings-emnlp.1006", pages = "15073--15079", abstract = "Produced in the form of small injections such as {``}Yeah!{''} or {``}Uh-Huh{''} by listeners in a conversation, supportive verbal feedback (i.e., backchanneling) is essential for natural dialogue. Highlighting its tight relation to speaker intent and utterance type, we propose a multi-task learning approach that learns textual representations for the task of backchannel prediction in tandem with dialogue act classification. We demonstrate the effectiveness of our approach by improving the prediction of specific backchannels like {``}Yeah{''} or {``}Really?{''} by up to 2.0{\%} in F1. Additionally, whereas previous models relied on well-established methods to extract audio features, we further pre-train the audio encoder in a self-supervised fashion using voice activity projection. This leads to additional gains of 1.4{\%} in weighted F1.", }

Produced in the form of small injections such as {``}Yeah!{''} or {``}Uh-Huh{''} by listeners in a conversation, supportive verbal feedback (i.e., backchanneling) is essential for natural dialogue. Highlighting its tight relation to speaker intent and utterance type, we propose a multi-task learning approach that learns textual representations for the task of backchannel prediction in tandem with dialogue act classification. We demonstrate the effectiveness of our approach by improving the prediction of specific backchannels like {``}Yeah{''} or {``}Really?{''} by up to 2.0{\%} in F1. Additionally, whereas previous models relied on well-established methods to extract audio features, we further pre-train the audio encoder in a self-supervised fashion using voice activity projection. This leads to additional gains of 1.4{\%} in weighted F1.

Dialogue Act-Aided Backchannel Prediction Using Multi-Task Learning

findings-emnlp.1006

Poster

https://aclanthology.org/2023.findings-emnlp.1007.bib

https://aclanthology.org/2023.findings-emnlp.1007/

@inproceedings{limkonchotiwat-etal-2023-mrefined, title = "m{R}e{F}in{ED}: An Efficient End-to-End Multilingual Entity Linking System", author = "Limkonchotiwat, Peerat and Cheng, Weiwei and Christodoulopoulos, Christos and Saffari, Amir and Lehmann, Jens", 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.1007", doi = "10.18653/v1/2023.findings-emnlp.1007", pages = "15080--15089", abstract = "End-to-end multilingual entity linking (MEL) is concerned with identifying multilingual entity mentions and their corresponding entity IDs in a knowledge base. Existing works assumed that entity mentions were given and skipped the entity mention detection step due to a lack of high-quality multilingual training corpora. To overcome this limitation, we propose mReFinED, the first end-to-end multilingual entity linking. Additionally, we propose a bootstrapping mention detection framework that enhances the quality of training corpora. Our experimental results demonstrated that mReFinED outperformed the best existing work in the end-to-end MEL task while being 44 times faster.", }

End-to-end multilingual entity linking (MEL) is concerned with identifying multilingual entity mentions and their corresponding entity IDs in a knowledge base. Existing works assumed that entity mentions were given and skipped the entity mention detection step due to a lack of high-quality multilingual training corpora. To overcome this limitation, we propose mReFinED, the first end-to-end multilingual entity linking. Additionally, we propose a bootstrapping mention detection framework that enhances the quality of training corpora. Our experimental results demonstrated that mReFinED outperformed the best existing work in the end-to-end MEL task while being 44 times faster.

mReFinED: An Efficient End-to-End Multilingual Entity Linking System

findings-emnlp.1007

Poster

https://aclanthology.org/2023.findings-emnlp.1008.bib

https://aclanthology.org/2023.findings-emnlp.1008/

@inproceedings{ke-etal-2023-sub, title = "Sub-network Discovery and Soft-masking for Continual Learning of Mixed Tasks", author = "Ke, Zixuan and Liu, Bing and Xiong, Wenhan and Celikyilmaz, Asli and Li, Haoran", 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.1008", doi = "10.18653/v1/2023.findings-emnlp.1008", pages = "15090--15107", abstract = "Continual learning (CL) has two main objectives: preventing catastrophic forgetting (CF) and encouraging knowledge transfer (KT). The existing literature mainly focused on overcoming CF. Some work has also been done on KT when the tasks are similar. To our knowledge, only one method has been proposed to learn a sequence of mixed tasks. However, these techniques still suffer from CF and/or limited KT. This paper proposes a new CL method to achieve both. It overcomes CF by isolating the knowledge of each task via discovering a sub-network for it. A soft-masking mechanism is also proposed to preserve the previous knowledge and to enable the new task to leverage the past knowledge to achieve KT. Experiments using classification, generation, information extraction, and their mixture (i.e., heterogeneous tasks) show that the proposed method consistently outperforms strong baselines.", }

Continual learning (CL) has two main objectives: preventing catastrophic forgetting (CF) and encouraging knowledge transfer (KT). The existing literature mainly focused on overcoming CF. Some work has also been done on KT when the tasks are similar. To our knowledge, only one method has been proposed to learn a sequence of mixed tasks. However, these techniques still suffer from CF and/or limited KT. This paper proposes a new CL method to achieve both. It overcomes CF by isolating the knowledge of each task via discovering a sub-network for it. A soft-masking mechanism is also proposed to preserve the previous knowledge and to enable the new task to leverage the past knowledge to achieve KT. Experiments using classification, generation, information extraction, and their mixture (i.e., heterogeneous tasks) show that the proposed method consistently outperforms strong baselines.

Sub-network Discovery and Soft-masking for Continual Learning of Mixed Tasks

findings-emnlp.1008

2310.09436

Poster

https://aclanthology.org/2023.findings-emnlp.1009.bib

https://aclanthology.org/2023.findings-emnlp.1009/

@inproceedings{lu-etal-2023-pivoine, title = "{PIVOINE}: Instruction Tuning for Open-world Entity Profiling", author = "Lu, Keming and Pan, Xiaoman and Song, Kaiqiang and Zhang, Hongming and Yu, Dong and Chen, Jianshu", 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.1009", doi = "10.18653/v1/2023.findings-emnlp.1009", pages = "15108--15127", abstract = "This work considers the problem of Open-world Entity Profiling, a sub-domain of Open-world Information Extraction (Open-world IE). Unlike the conventional closed-world IE, Open-world IE is considered a more general situation where entities and relations could be beyond a predefined ontology. We seek to develop a large language model (LLM) that can perform Open-world Entity Profiling with instruction tuning to extract desirable entity profiles characterized by (possibly fine-grained) natural language instructions. In particular, we construct INSTRUCTOPENWIKI, a substantial instruction-tuning dataset for Open-world Entity Profiling enriched with a comprehensive corpus, extensive annotations, and diverse instructions. We finetune pretrained BLOOM models on INSTRUCTOPENWIKI and obtain PIVOINE, an LLM for Open-world Entity Profiling with strong instruction-following capabilities. Our experiments demonstrate that PIVOINE significantly outperforms traditional methods and ChatGPT-based baselines, displaying impressive generalization capabilities on both unseen instructions and out-of-ontology cases. Consequently, PIVOINE emerges as a promising solution to tackle the open-world challenge of entity profiling.", }

This work considers the problem of Open-world Entity Profiling, a sub-domain of Open-world Information Extraction (Open-world IE). Unlike the conventional closed-world IE, Open-world IE is considered a more general situation where entities and relations could be beyond a predefined ontology. We seek to develop a large language model (LLM) that can perform Open-world Entity Profiling with instruction tuning to extract desirable entity profiles characterized by (possibly fine-grained) natural language instructions. In particular, we construct INSTRUCTOPENWIKI, a substantial instruction-tuning dataset for Open-world Entity Profiling enriched with a comprehensive corpus, extensive annotations, and diverse instructions. We finetune pretrained BLOOM models on INSTRUCTOPENWIKI and obtain PIVOINE, an LLM for Open-world Entity Profiling with strong instruction-following capabilities. Our experiments demonstrate that PIVOINE significantly outperforms traditional methods and ChatGPT-based baselines, displaying impressive generalization capabilities on both unseen instructions and out-of-ontology cases. Consequently, PIVOINE emerges as a promising solution to tackle the open-world challenge of entity profiling.

PIVOINE: Instruction Tuning for Open-world Entity Profiling

findings-emnlp.1009

Poster

https://aclanthology.org/2023.findings-emnlp.1010.bib

https://aclanthology.org/2023.findings-emnlp.1010/

@inproceedings{zhao-etal-2023-diqad, title = "{D}i{QAD}: A Benchmark Dataset for Open-domain Dialogue Quality Assessment", author = "Zhao, Yukun and Yan, Lingyong and Sun, Weiwei and Meng, Chong and Wang, Shuaiqiang and Cheng, Zhicong and Ren, Zhaochun and Yin, Dawei", 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.1010", doi = "10.18653/v1/2023.findings-emnlp.1010", pages = "15128--15145", abstract = "Dialogue assessment plays a critical role in the development of open-domain dialogue systems. Existing work are uncapable of providing an end-to-end and human-epistemic assessment dataset, while they only provide sub-metrics like coherence or the dialogues are conversed between annotators far from real user settings. In this paper, we release a large-scale dialogue quality assessment dataset (DiQAD), for automatically assessing open-domain dialogue quality. Specifically, we (1) establish the assessment criteria based on the dimensions conforming to human judgements on dialogue qualities, and (2) annotate large-scale dialogues that conversed between real users based on these annotation criteria, which contains around 100,000 dialogues. We conduct several experiments and report the performances of the baselines as the benchmark on DiQAD. The dataset is openly accessible at \url{https://github.com/yukunZhao/Dataset_Dialogue_quality_evaluation}.", }

Dialogue assessment plays a critical role in the development of open-domain dialogue systems. Existing work are uncapable of providing an end-to-end and human-epistemic assessment dataset, while they only provide sub-metrics like coherence or the dialogues are conversed between annotators far from real user settings. In this paper, we release a large-scale dialogue quality assessment dataset (DiQAD), for automatically assessing open-domain dialogue quality. Specifically, we (1) establish the assessment criteria based on the dimensions conforming to human judgements on dialogue qualities, and (2) annotate large-scale dialogues that conversed between real users based on these annotation criteria, which contains around 100,000 dialogues. We conduct several experiments and report the performances of the baselines as the benchmark on DiQAD. The dataset is openly accessible at \url{https://github.com/yukunZhao/Dataset_Dialogue_quality_evaluation}.

DiQAD: A Benchmark Dataset for Open-domain Dialogue Quality Assessment

findings-emnlp.1010

Poster

https://aclanthology.org/2023.findings-emnlp.1011.bib

https://aclanthology.org/2023.findings-emnlp.1011/

@inproceedings{li-etal-2023-tuna, title = "Tuna: Instruction Tuning using Feedback from Large Language Models", author = "Li, Haoran and Liu, Yiran and Zhang, Xingxing and Lu, Wei and Wei, Furu", 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.1011", doi = "10.18653/v1/2023.findings-emnlp.1011", pages = "15146--15163", abstract = "Instruction tuning of open-source large language models (LLMs) like LLaMA, using direct outputs from more powerful LLMs such as Instruct-GPT and GPT-4, has proven to be a cost-effective way to align model behaviors with human preferences. However, the instruction-tuned model has only seen one response per instruction, lacking the knowledge of potentially better responses. In this paper, we propose finetuning an instruction-tuned LLM using our novel probabilistic ranking and contextual ranking approaches to increase the likelihood of generating better responses. Probabilistic ranking enables the instruction-tuned model to inherit the relative rankings of high-quality and low-quality responses from the teacher LLM. On the other hand, learning with contextual ranking allows the model to refine its own response distribution using the contextual understanding ability of stronger LLMs. Furthermore, we apply probabilistic ranking and contextual ranking sequentially to the instruction-tuned LLM. The resulting model, which we call Tuna, consistently improves the performance on Super Natural Instructions (119 test tasks), LMentry (25 test tasks), Vicuna QA, and can even obtain better results than several strong reinforcement learning baselines. Our code and data are available at https://github.com/microsoft/LMOps.", }

Instruction tuning of open-source large language models (LLMs) like LLaMA, using direct outputs from more powerful LLMs such as Instruct-GPT and GPT-4, has proven to be a cost-effective way to align model behaviors with human preferences. However, the instruction-tuned model has only seen one response per instruction, lacking the knowledge of potentially better responses. In this paper, we propose finetuning an instruction-tuned LLM using our novel probabilistic ranking and contextual ranking approaches to increase the likelihood of generating better responses. Probabilistic ranking enables the instruction-tuned model to inherit the relative rankings of high-quality and low-quality responses from the teacher LLM. On the other hand, learning with contextual ranking allows the model to refine its own response distribution using the contextual understanding ability of stronger LLMs. Furthermore, we apply probabilistic ranking and contextual ranking sequentially to the instruction-tuned LLM. The resulting model, which we call Tuna, consistently improves the performance on Super Natural Instructions (119 test tasks), LMentry (25 test tasks), Vicuna QA, and can even obtain better results than several strong reinforcement learning baselines. Our code and data are available at https://github.com/microsoft/LMOps.

Tuna: Instruction Tuning using Feedback from Large Language Models

findings-emnlp.1011

2310.13385

https://huggingface.co/papers/2310.13385

Poster

https://aclanthology.org/2023.findings-emnlp.1012.bib

https://aclanthology.org/2023.findings-emnlp.1012/

@inproceedings{pinter-elhadad-2023-emptying, title = "Emptying the Ocean with a Spoon: Should We Edit Models?", author = "Pinter, Yuval and Elhadad, Michael", 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.1012", doi = "10.18653/v1/2023.findings-emnlp.1012", pages = "15164--15172", abstract = "We call into question the recently popularized method of direct model editing as a means of correcting factual errors in LLM generations. We contrast model editing with three similar but distinct approaches that pursue better defined objectives: (1) retrieval-based architectures, which decouple factual memory from inference and linguistic capabilities embodied in LLMs; (2) concept erasure methods, which aim at preventing systemic bias in generated text; and (3) attribution methods, which aim at grounding generations into identified textual sources. We argue that direct model editing cannot be trusted as a systematic remedy for the disadvantages inherent to LLMs, and while it has proven potential in improving model explainability, it opens risks by reinforcing the notion that models can be trusted for factuality. We call for cautious promotion and application of model editing as part of the LLM deployment process, and for responsibly limiting the use cases of LLMs to those not relying on editing as a critical component.", }

We call into question the recently popularized method of direct model editing as a means of correcting factual errors in LLM generations. We contrast model editing with three similar but distinct approaches that pursue better defined objectives: (1) retrieval-based architectures, which decouple factual memory from inference and linguistic capabilities embodied in LLMs; (2) concept erasure methods, which aim at preventing systemic bias in generated text; and (3) attribution methods, which aim at grounding generations into identified textual sources. We argue that direct model editing cannot be trusted as a systematic remedy for the disadvantages inherent to LLMs, and while it has proven potential in improving model explainability, it opens risks by reinforcing the notion that models can be trusted for factuality. We call for cautious promotion and application of model editing as part of the LLM deployment process, and for responsibly limiting the use cases of LLMs to those not relying on editing as a critical component.

Emptying the Ocean with a Spoon: Should We Edit Models?

findings-emnlp.1012

2310.11958

Poster

https://aclanthology.org/2023.findings-emnlp.1013.bib

https://aclanthology.org/2023.findings-emnlp.1013/

@inproceedings{wang-etal-2023-causal, title = "A Causal View of Entity Bias in (Large) Language Models", author = "Wang, Fei and Mo, Wenjie and Wang, Yiwei and Zhou, Wenxuan and Chen, Muhao", 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.1013", doi = "10.18653/v1/2023.findings-emnlp.1013", pages = "15173--15184", abstract = "Entity bias widely affects pretrained (large) language models, causing them to rely on (biased) parametric knowledge to make unfaithful predictions. Although causality-inspired methods have shown great potential to mitigate entity bias, it is hard to precisely estimate the parameters of underlying causal models in practice. The rise of black-box LLMs also makes the situation even worse, because of their inaccessible parameters and uncalibrated logits. To address these problems, we propose a specific structured causal model (SCM) whose parameters are comparatively easier to estimate. Building upon this SCM, we propose causal intervention techniques to mitigate entity bias for both white-box and black-box settings. The proposed causal intervention perturbs the original entity with neighboring entities. This intervention reduces specific biasing information pertaining to the original entity while still preserving sufficient semantic information from similar entities. Under the white-box setting, our training-time intervention improves OOD performance of PLMs on relation extraction (RE) and machine reading comprehension (MRC) by 5.7 points and by 9.1 points, respectively. Under the black-box setting, our in-context intervention effectively reduces the entity-based knowledge conflicts of GPT-3.5, achieving up to 20.5 points of improvement of exact match accuracy on MRC and up to 17.6 points of reduction in memorization ratio on RE.", }

Entity bias widely affects pretrained (large) language models, causing them to rely on (biased) parametric knowledge to make unfaithful predictions. Although causality-inspired methods have shown great potential to mitigate entity bias, it is hard to precisely estimate the parameters of underlying causal models in practice. The rise of black-box LLMs also makes the situation even worse, because of their inaccessible parameters and uncalibrated logits. To address these problems, we propose a specific structured causal model (SCM) whose parameters are comparatively easier to estimate. Building upon this SCM, we propose causal intervention techniques to mitigate entity bias for both white-box and black-box settings. The proposed causal intervention perturbs the original entity with neighboring entities. This intervention reduces specific biasing information pertaining to the original entity while still preserving sufficient semantic information from similar entities. Under the white-box setting, our training-time intervention improves OOD performance of PLMs on relation extraction (RE) and machine reading comprehension (MRC) by 5.7 points and by 9.1 points, respectively. Under the black-box setting, our in-context intervention effectively reduces the entity-based knowledge conflicts of GPT-3.5, achieving up to 20.5 points of improvement of exact match accuracy on MRC and up to 17.6 points of reduction in memorization ratio on RE.

A Causal View of Entity Bias in (Large) Language Models

findings-emnlp.1013

2305.14695

Poster

https://aclanthology.org/2023.findings-emnlp.1014.bib

https://aclanthology.org/2023.findings-emnlp.1014/

@inproceedings{qin-etal-2023-t5score, title = "{T}5{S}core: Discriminative Fine-tuning of Generative Evaluation Metrics", author = "Qin, Yiwei and Yuan, Weizhe and Neubig, Graham and Liu, Pengfei", 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.1014", doi = "10.18653/v1/2023.findings-emnlp.1014", pages = "15185--15202", abstract = "Modern embedding-based metrics for evaluation of generated text generally fall into one of two paradigms: discriminative metrics that are trained to directly predict which outputs are of higher quality according to supervised human annotations, and generative metrics that are trained to evaluate text based on the probabilities of a generative model. Both have their advantages; discriminative metrics are able to directly optimize for the problem of distinguishing between good and bad outputs, while generative metrics can be trained using abundant raw text. In this paper, we present a framework that combines the best of both worlds, using both supervised and unsupervised signals from whatever data we have available. We operationalize this idea by training T5Score, a metric that uses these training signals with mT5 as backbone. We perform an extensive empirical comparison with other existing metrics on 5 datasets, 19 languages and 280 systems, demonstrating the utility of our method. Experimental results show that: T5Score achieves the best performance on all datasets against existing top-scoring metrics at the segment level.", }

Modern embedding-based metrics for evaluation of generated text generally fall into one of two paradigms: discriminative metrics that are trained to directly predict which outputs are of higher quality according to supervised human annotations, and generative metrics that are trained to evaluate text based on the probabilities of a generative model. Both have their advantages; discriminative metrics are able to directly optimize for the problem of distinguishing between good and bad outputs, while generative metrics can be trained using abundant raw text. In this paper, we present a framework that combines the best of both worlds, using both supervised and unsupervised signals from whatever data we have available. We operationalize this idea by training T5Score, a metric that uses these training signals with mT5 as backbone. We perform an extensive empirical comparison with other existing metrics on 5 datasets, 19 languages and 280 systems, demonstrating the utility of our method. Experimental results show that: T5Score achieves the best performance on all datasets against existing top-scoring metrics at the segment level.

T5Score: Discriminative Fine-tuning of Generative Evaluation Metrics

findings-emnlp.1014

2212.05726

Poster

https://aclanthology.org/2023.findings-emnlp.1015.bib

https://aclanthology.org/2023.findings-emnlp.1015/

@inproceedings{garcia-ferrero-etal-2023-projection, title = "{T}-Projection: High Quality Annotation Projection for Sequence Labeling Tasks", author = "Garc{\'\i}a-Ferrero, Iker and Agerri, Rodrigo and Rigau, German", 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.1015", doi = "10.18653/v1/2023.findings-emnlp.1015", pages = "15203--15217", abstract = "In the absence of readily available labeled data for a given sequence labeling task and language, annotation projection has been proposed as one of the possible strategies to automatically generate annotated data. Annotation projection has often been formulated as the task of transporting, on parallel corpora, the labels pertaining to a given span in the source language into its corresponding span in the target language. In this paper we present T-Projection, a novel approach for annotation projection that leverages large pretrained text2text language models and state-of-the-art machine translation technology. T-Projection decomposes the label projection task into two subtasks: (i) A candidate generation step, in which a set of projection candidates using a multilingual T5 model is generated and, (ii) a candidate selection step, in which the generated candidates are ranked based on translation probabilities. We conducted experiments on intrinsic and extrinsic tasks in 5 Indo-European and 8 low-resource African languages. We demostrate that T-projection outperforms previous annotation projection methods by a wide margin. We believe that T-Projection can help to automatically alleviate the lack of high-quality training data for sequence labeling tasks. Code and data are publicly available.", }

In the absence of readily available labeled data for a given sequence labeling task and language, annotation projection has been proposed as one of the possible strategies to automatically generate annotated data. Annotation projection has often been formulated as the task of transporting, on parallel corpora, the labels pertaining to a given span in the source language into its corresponding span in the target language. In this paper we present T-Projection, a novel approach for annotation projection that leverages large pretrained text2text language models and state-of-the-art machine translation technology. T-Projection decomposes the label projection task into two subtasks: (i) A candidate generation step, in which a set of projection candidates using a multilingual T5 model is generated and, (ii) a candidate selection step, in which the generated candidates are ranked based on translation probabilities. We conducted experiments on intrinsic and extrinsic tasks in 5 Indo-European and 8 low-resource African languages. We demostrate that T-projection outperforms previous annotation projection methods by a wide margin. We believe that T-Projection can help to automatically alleviate the lack of high-quality training data for sequence labeling tasks. Code and data are publicly available.

T-Projection: High Quality Annotation Projection for Sequence Labeling Tasks

findings-emnlp.1015

2212.10548

https://huggingface.co/papers/2212.10548

Poster

https://aclanthology.org/2023.findings-emnlp.1016.bib

https://aclanthology.org/2023.findings-emnlp.1016/

@inproceedings{chu-etal-2023-mtger, title = "{MTGER}: Multi-view Temporal Graph Enhanced Temporal Reasoning over Time-Involved Document", author = "Chu, Zheng and Wang, Zekun and Liang, Jiafeng and Liu, Ming and Qin, Bing", 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.1016", doi = "10.18653/v1/2023.findings-emnlp.1016", pages = "15218--15233", abstract = "The facts and time in the document are intricately intertwined, making temporal reasoning over documents challenging. Previous work models time implicitly, making it difficult to handle such complex relationships. To address this issue, we propose MTGER, a novel Multi-view Temporal Graph Enhanced Reasoning framework for temporal reasoning over time-involved documents. Concretely, MTGER explicitly models the temporal relationships among facts by multi-view temporal graphs. On the one hand, the heterogeneous temporal graphs explicitly model the temporal and discourse relationships among facts; on the other hand, the multi-view mechanism captures both time-focused and fact-focused information, allowing the two views to complement each other through adaptive fusion. To further improve the implicit reasoning capability of the model, we design a self-supervised time-comparing objective. Extensive experimental results demonstrate the effectiveness of our method on the TimeQA and SituatedQA datasets. Furthermore, MTGER gives more consistent answers under question perturbations.", }

The facts and time in the document are intricately intertwined, making temporal reasoning over documents challenging. Previous work models time implicitly, making it difficult to handle such complex relationships. To address this issue, we propose MTGER, a novel Multi-view Temporal Graph Enhanced Reasoning framework for temporal reasoning over time-involved documents. Concretely, MTGER explicitly models the temporal relationships among facts by multi-view temporal graphs. On the one hand, the heterogeneous temporal graphs explicitly model the temporal and discourse relationships among facts; on the other hand, the multi-view mechanism captures both time-focused and fact-focused information, allowing the two views to complement each other through adaptive fusion. To further improve the implicit reasoning capability of the model, we design a self-supervised time-comparing objective. Extensive experimental results demonstrate the effectiveness of our method on the TimeQA and SituatedQA datasets. Furthermore, MTGER gives more consistent answers under question perturbations.

MTGER: Multi-view Temporal Graph Enhanced Temporal Reasoning over Time-Involved Document

findings-emnlp.1016

2311.04816

Poster

https://aclanthology.org/2023.findings-emnlp.1017.bib

https://aclanthology.org/2023.findings-emnlp.1017/

@inproceedings{dongge-yang-2023-mscffn, title = "{MSCFFN}: A New {FFN} with Multi-Space Cross to Accelerate Transformer", author = "Dongge, Tang and Yang, Qing", 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.1017", doi = "10.18653/v1/2023.findings-emnlp.1017", pages = "15234--15239", abstract = "Transformer models have achieved impressive success in various natural language processing tasks. But it is also limited used in some areas and the heavy computation complexity is one of the main limitations. Many model structures have been proposed to reduce the computation complexity and some are really effective. The previous research can be divided into two categories. One is to use more effective training and inference strategies and the other is focused on how to replace the standard self-attention mechanism with linear attention method. Differently, we revisit the design in Transformer and find that the feed forward network (FFN) is also computationally expensive, especially when the hidden dimension is large. In this paper, we propose a new FFN structure, named MSCFFN, which splits the large matrix space to several small space to reduce the computation complexity and uses the Multi-Space Cross method to ensure the accurate result. To the best of our knowledge, this is the first time to redesign FFN to accelerate Transformers. We experimentally validate the effectiveness of the proposed method on the Long-Range Arena benchmark. And the results show MSCFFN can achieve a faster speed with a similar or even better accuracy.", }

Transformer models have achieved impressive success in various natural language processing tasks. But it is also limited used in some areas and the heavy computation complexity is one of the main limitations. Many model structures have been proposed to reduce the computation complexity and some are really effective. The previous research can be divided into two categories. One is to use more effective training and inference strategies and the other is focused on how to replace the standard self-attention mechanism with linear attention method. Differently, we revisit the design in Transformer and find that the feed forward network (FFN) is also computationally expensive, especially when the hidden dimension is large. In this paper, we propose a new FFN structure, named MSCFFN, which splits the large matrix space to several small space to reduce the computation complexity and uses the Multi-Space Cross method to ensure the accurate result. To the best of our knowledge, this is the first time to redesign FFN to accelerate Transformers. We experimentally validate the effectiveness of the proposed method on the Long-Range Arena benchmark. And the results show MSCFFN can achieve a faster speed with a similar or even better accuracy.

MSCFFN: A New FFN with Multi-Space Cross to Accelerate Transformer

findings-emnlp.1017

Poster

https://aclanthology.org/2023.findings-emnlp.1018.bib

https://aclanthology.org/2023.findings-emnlp.1018/

@inproceedings{paonessa-etal-2023-dialect, title = "Dialect Transfer for {S}wiss {G}erman Speech Translation", author = {Paonessa, Claudio and Schraner, Yanick and Deriu, Jan and H{\"u}rlimann, Manuela and Vogel, Manfred and Cieliebak, Mark}, 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.1018", doi = "10.18653/v1/2023.findings-emnlp.1018", pages = "15240--15254", abstract = "This paper investigates the challenges in building Swiss German speech translation systems, specifically focusing on the impact of dialect diversity and differences between Swiss German and Standard German. Swiss German is a spoken language with no formal writing system, it comprises many diverse dialects and is a low-resource language with only around 5 million speakers. The study is guided by two key research questions: how does the inclusion and exclusion of dialects during the training of speech translation models for Swiss German impact the performance on specific dialects, and how do the differences between Swiss German and Standard German impact the performance of the systems? We show that dialect diversity and linguistic differences pose significant challenges to Swiss German speech translation, which is in line with linguistic hypotheses derived from empirical investigations.", }

This paper investigates the challenges in building Swiss German speech translation systems, specifically focusing on the impact of dialect diversity and differences between Swiss German and Standard German. Swiss German is a spoken language with no formal writing system, it comprises many diverse dialects and is a low-resource language with only around 5 million speakers. The study is guided by two key research questions: how does the inclusion and exclusion of dialects during the training of speech translation models for Swiss German impact the performance on specific dialects, and how do the differences between Swiss German and Standard German impact the performance of the systems? We show that dialect diversity and linguistic differences pose significant challenges to Swiss German speech translation, which is in line with linguistic hypotheses derived from empirical investigations.

Dialect Transfer for Swiss German Speech Translation

findings-emnlp.1018

2310.09088

Poster

https://aclanthology.org/2023.findings-emnlp.1019.bib

https://aclanthology.org/2023.findings-emnlp.1019/

@inproceedings{dou-etal-2023-masked, title = "Masked Path Modeling for Vision-and-Language Navigation", author = "Dou, Zi-Yi and Gao, Feng and Peng, Nanyun", 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.1019", doi = "10.18653/v1/2023.findings-emnlp.1019", pages = "15255--15269", abstract = "Vision-and-language navigation (VLN) agents are trained to navigate in real-world environments based on natural language instructions. A major challenge in VLN is the limited available training data, which hinders the models{'} ability to generalize effectively. Previous approaches have attempted to alleviate this issue by using external tools to generate pseudo-labeled data or integrating web-scaled image-text pairs during training. However, these methods often rely on automatically-generated or out-of-domain data, leading to challenges such as suboptimal data quality and domain mismatch. In this paper, we introduce a masked path modeling (MPM) objective. MPM pretrains an agent using self-collected data for subsequent navigation tasks, eliminating the need for external tools. Specifically, our method allows the agent to explore navigation environments and record the paths it traverses alongside the corresponding agent actions. Subsequently, we train the agent on this collected data to reconstruct the original action sequence when given a randomly masked subsequence of the original path. This approach enables the agent to accumulate a diverse and substantial dataset, facilitating the connection between visual observations of paths and the agent{'}s actions, which is the foundation of the VLN task. Importantly, the collected data are in-domain, and the training process avoids synthetic data with uncertain quality, addressing previous issues. We conduct experiments on various VLN datasets and demonstrate the applications of MPM across different levels of instruction complexity. Our results exhibit significant improvements in success rates, with enhancements of 1.3{\%}, 1.1{\%}, and 1.2{\%} on the val-unseen split of the Room-to-Room, Room-for-Room, and Room-across-Room datasets, respectively. Additionally, we underscore the adaptability of MPM as well as the potential for additional improvements when the agent is allowed to explore unseen environments prior to testing.", }

Vision-and-language navigation (VLN) agents are trained to navigate in real-world environments based on natural language instructions. A major challenge in VLN is the limited available training data, which hinders the models{'} ability to generalize effectively. Previous approaches have attempted to alleviate this issue by using external tools to generate pseudo-labeled data or integrating web-scaled image-text pairs during training. However, these methods often rely on automatically-generated or out-of-domain data, leading to challenges such as suboptimal data quality and domain mismatch. In this paper, we introduce a masked path modeling (MPM) objective. MPM pretrains an agent using self-collected data for subsequent navigation tasks, eliminating the need for external tools. Specifically, our method allows the agent to explore navigation environments and record the paths it traverses alongside the corresponding agent actions. Subsequently, we train the agent on this collected data to reconstruct the original action sequence when given a randomly masked subsequence of the original path. This approach enables the agent to accumulate a diverse and substantial dataset, facilitating the connection between visual observations of paths and the agent{'}s actions, which is the foundation of the VLN task. Importantly, the collected data are in-domain, and the training process avoids synthetic data with uncertain quality, addressing previous issues. We conduct experiments on various VLN datasets and demonstrate the applications of MPM across different levels of instruction complexity. Our results exhibit significant improvements in success rates, with enhancements of 1.3{\%}, 1.1{\%}, and 1.2{\%} on the val-unseen split of the Room-to-Room, Room-for-Room, and Room-across-Room datasets, respectively. Additionally, we underscore the adaptability of MPM as well as the potential for additional improvements when the agent is allowed to explore unseen environments prior to testing.

Masked Path Modeling for Vision-and-Language Navigation

findings-emnlp.1019

2305.14268

Poster