EMNLP 2023 论文列表

Prompt-based learning has emerged as a powerful technique in natural language processing (NLP) due to its ability to leverage pre-training knowledge for downstream few-shot tasks. In this paper, we propose 2INER, a novel text-to-text framework for Few-Shot Named Entity Recognition (NER) tasks. Our approach employs instruction finetuning based on InstructionNER to enable the model to effectively comprehend and process task-specific instructions, including both main and auxiliary tasks. We also introduce a new auxiliary task, called Type Extracting, to enhance the model's understanding of entity types in the overall semantic context of a sentence. To facilitate in-context learning, we concatenate examples to the input, enabling the model to learn from additional contextual information. Experimental results on four datasets demonstrate that our approach outperforms existing Few-Shot NER methods and remains competitive with state-of-the-art standard NER algorithms.

Emotion Cause Triplet Extraction in Conversations (ECTEC) aims to simultaneously extract emotion utterances, emotion categories, and cause utterances from conversations. However, existing studies mainly decompose the ECTEC task into multiple subtasks and solve them in a pipeline manner. Moreover, since conversations tend to contain many informal and implicit expressions, it often requires external knowledge and reasoning-based inference to accurately identify emotional and causal clues implicitly mentioned in the context, which are ignored by previous work. To address these limitations, in this paper, we propose a commonSense knowledge-enHanced generAtive fRameworK named SHARK, which formulates the ECTEC task as an index generation problem and generates the emotion-cause-category triplets in an end-to-end manner with a sequence-to-sequence model. Furthermore, we propose to incorporate both retrieved and generated commonsense knowledge into the generative model via a dual-view gate mechanism and a graph attention layer. Experimental results show that our SHARK model consistently outperforms several competitive systems on two benchmark datasets. Our source codes are publicly released at https://github.com/NUSTM/SHARK.

Large Language Models (LLMs) have significantly advanced the field of Natural Language Processing (NLP), but their lack of interpretability has been a major concern. Current methods for interpreting LLMs are post hoc, applied after inference time, and have limitations such as their focus on low-level features and lack of explainability at higher-level text units. In this work, we introduce proto-lm, a prototypical network-based white-box framework that allows LLMs to learn immediately interpretable embeddings during the fine-tuning stage while maintaining competitive performance. Our method's applicability and interpretability are demonstrated through experiments on a wide range of NLP tasks, and our results indicate a new possibility of creating interpretable models without sacrificing performance. This novel approach to interpretability in LLMs can pave the way for more interpretable models without the need to sacrifice performance. We release our code at https://github.com/yx131/proto-lm}{https://github.com/yx131/proto-lm.

Cross-lingual transfer learning heavily relies on well-aligned cross-lingual representations. The syntactic structure is recognized as beneficial for cross-lingual transfer, but limited researches utilize it for aligning representation in multilingual pre-trained language models (PLMs). Additionally, existing methods require syntactic labels that are difficult to obtain and of poor quality for low-resource languages. To address this gap, we propose Struct-XLM, a novel multilingual language model that leverages reinforcement learning (RL) to autonomously discover universal syntactic structures for improving the cross-lingual representation alignment of PLM. Struct-XLM integrates a policy network (PNet) and a translation ranking task. The PNet is designed to discover structural information and integrate it into the last layer of the PLM through the structural multi-head attention module to obtain structural representation. The translation ranking task obtains a delayed reward based on the structural representation to optimize the PNet while improving the alignment of cross-lingual representation. Experiments show the effectiveness of the proposed approach for enhancing cross-lingual transfer of multilingual PLM on the XTREME benchmark.

Recent work has found that few-shot sentence classification based on pre-trained Sentence Encoders (SEs) is efficient, robust, and effective. In this work, we investigate strategies for domain-specialization in the context of few-shot sentence classification with SEs. We first establish that unsupervised Domain-Adaptive Pre-Training (DAPT) of a base Pre-trained Language Model (PLM) (i.e., not an SE) substantially improves the accuracy of few-shot sentence classification by up to 8.4 points. However, applying DAPT on SEs, on the one hand, disrupts the effects of their (general-domain) Sentence Embedding Pre-Training (SEPT). On the other hand, applying general-domain SEPT on top of a domain-adapted base PLM (i.e., after DAPT) is effective but inefficient, since the computationally expensive SEPT needs to be executed on top of a DAPT-ed PLM of each domain. As a solution, we propose AdaSent, which decouples SEPT from DAPT by training a SEPT adapter on the base PLM. The adapter can be inserted into DAPT-ed PLMs from any domain. We demonstrate AdaSent's effectiveness in extensive experiments on 17 different few-shot sentence classification datasets. AdaSent matches or surpasses the performance of full SEPT on DAPT-ed PLM, while substantially reducing the training costs. The code for AdaSent is available.

Conversational Question Answering (CQA) aims to provide natural language answers to users in information-seeking dialogues. Existing CQA benchmarks often evaluate models using pre-collected human-human conversations. However, replacing the model-predicted dialogue history with ground truth compromises the naturalness and sustainability of CQA evaluation. While previous studies proposed using predicted history and rewriting techniques to address unresolved coreferences and incoherencies, this approach renders the question self-contained from the conversation. In this paper, we propose a novel automatic evaluation approach, interview evaluation. Specifically, ChatGPT acts as the interviewer (Q agent) with a set of carefully designed prompts, and the CQA model under test serves as the interviewee (A agent). During the interview evaluation, questions are dynamically generated by the Q agent to guide the A agent in predicting the correct answer through an interactive process. We evaluated four different models on QuAC and two models on CoQA in our experiments. The experiment results demonstrate that our interview evaluation has advantages over previous CQA evaluation approaches, particularly in terms of naturalness and coherence. The source code is made publicly available.

Automated Essay Scoring (AES) aims to automatically assess the quality of essays. Automation enables large-scale assessment, improvements in consistency, reliability, and standardization. Those characteristics are of particular relevance in the context of language certification exams. However, a major bottleneck in the development of AES systems is the availability of corpora, which, unfortunately, are scarce, especially for languages other than English. In this paper, we aim to foster the development of AES for French by providing the TCFLE-8 corpus, a corpus of 6.5k essays collected in the context of the \textit{Test de Connaissance du Français} (TCF - French Knowledge Test) certification exam. We report the strict quality procedure that led to the scoring of each essay by at least two raters according to the CEFR levels and to the creation of a balanced corpus. In addition, we describe how linguistic properties of the essays relate to the learners' proficiency in TCFLE-8. We also advance the state-of-the-art performance for the AES task in French by experimenting with two strong baselines (i.e. RoBERTa and feature-based). Finally, we discuss the challenges of AES using TCFLE-8.

Large language models (e.g., GPT-4) are uniquely capable of producing highly rated text simplification, yet current human evaluation methods fail to provide a clear understanding of systems' specific strengths and weaknesses. To address this limitation, we introduce SALSA, an edit-based human annotation framework that enables holistic and fine-grained text simplification evaluation. We develop twenty one linguistically grounded edit types, covering the full spectrum of success and failure across dimensions of conceptual, syntactic and lexical simplicity. Using SALSA, we collect 19K edit annotations on 840 simplifications, revealing discrepancies in the distribution of simplification strategies performed by fine-tuned models, prompted LLMs and humans, and find GPT-3.5 performs more quality edits than humans, but still exhibits frequent errors. Using our fine-grained annotations, we develop LENS-SALSA, a reference-free automatic simplification metric, trained to predict sentence- and word-level quality simultaneously. Additionally, we introduce word-level quality estimation for simplification and report promising baseline results. Our data, new metric, and annotation toolkit are available at https://salsa-eval.com.

Research in the field of NLP has recently focused on the variability that people show in selecting labels when performing an annotation task. Exploiting disagreements in annotations has been shown to offer advantages for accurate modelling and fair evaluation. In this paper, we propose a strongly perspectivist model for supervised classification of natural language utterances. Our approach combines the predictions of several perspective-aware models using key information of their individual confidence to capture the subjectivity encoded in the annotation of linguistic phenomena. We validate our method through experiments on two case studies, irony and hate speech detection, in in-domain and cross-domain settings. The results show that confidence-based ensembling of perspective-aware models seems beneficial for classification performance in all scenarios. In addition, we demonstrate the effectiveness of our method with automatically extracted perspectives from annotations when the annotators' metadata are not available.

Recent large-scale Visual-Language Generative Models (VLGMs) have achieved unprecedented improvement in multimodal image/text generation. However, these models might also generate toxic content, e.g., offensive text and pornography images, raising significant ethical risks. Despite exhaustive studies on toxic degeneration of language models, this problem remains largely unexplored within the context of visual-language generation. This work delves into the propensity for toxicity generation and susceptibility to toxic data across various VLGMs. For this purpose, we built ToViLaG, a dataset comprising 32K co-toxic/mono-toxic text-image pairs and 1K innocuous but evocative text that tends to stimulate toxicity. Furthermore, we propose WInToRe, a novel toxicity metric tailored to visual-language generation, which theoretically reflects different aspects of toxicity considering both input and output. On such a basis, we benchmarked the toxicity of a diverse spectrum of VLGMs and discovered that some models do more evil than expected while some are more vulnerable to infection, underscoring the necessity of VLGMs detoxification. Therefore, we develop an innovative bottleneck-based detoxification method. Our method could reduce toxicity while maintaining comparable generation quality, providing a promising initial solution to this line of research.

In spite of the potential for ground-breaking achievements offered by large language models (LLMs) (e.g., GPT-3) via in-context learning (ICL), they still lag significantly behind fully-supervised baselines (e.g., fine-tuned BERT) in relation extraction (RE). This is due to the two major shortcomings of ICL for RE: (1) low relevance regarding entity and relation in existing sentence-level demonstration retrieval approaches for ICL; and (2) the lack of explaining input-label mappings of demonstrations leading to poor ICL effectiveness. In this paper, we propose GPT-RE to successfully address the aforementioned issues by (1) incorporating task-aware representations in demonstration retrieval; and (2) enriching the demonstrations with gold label-induced reasoning logic. We evaluate GPT-RE on four widely-used RE datasets, and observe that GPT-RE achieves improvements over not only existing GPT-3 baselines, but also fully-supervised baselines as in Figure 1. Specifically, GPT-RE achieves SOTA performances on the Semeval and SciERC datasets, and competitive performances on the TACRED and ACE05 datasets. Additionally, a critical issue of LLMs revealed by previous work, the strong inclination to wrongly classify NULL examples into other pre-defined labels, is substantially alleviated by our method. We show an empirical analysis.

Conditional story generation is significant in human-machine interaction, particularly in producing stories with complex plots. While Large language models (LLMs) perform well on multiple NLP tasks, including story generation, it is challenging to generate stories with both complex and creative plots. Existing methods often rely on detailed prompts to guide LLMs to meet target conditions, which inadvertently restrict the creative potential of the generated stories. We argue that leveraging information from exemplary human-written stories facilitates generating more diverse plotlines. Delving deeper into story details helps build complex and credible plots. In this paper, we propose a retrieval-auGmented stoRy generation framework with a fOrest of eVidEnce (GROVE) to enhance stories' complexity. We build a retrieval repository for target conditions to produce few-shot examples to prompt LLMs. Additionally, we design an "asking-why" prompting scheme that extracts a forest of evidence, providing compensation for the ambiguities that may occur in the generated story. This iterative process uncovers underlying story backgrounds. Finally, we select the most fitting chains of evidence from the evidence forest and integrate them into the generated story, thereby enhancing the narrative's complexity and credibility. Experimental results and numerous examples verify the effectiveness of our method.

Social media has not only facilitated news consumption, but also led to the wide spread of fake news. Because news articles in social media is usually condensed and full of knowledge entities, existing methods of fake news detection use external entity knowledge. However, majority of these methods focus on news entity information and ignore the structured knowledge among news entities. To address this issue, in this work, we propose a Knowledge grAPh enhAnced Language Model (KAPALM) which is a novel model that fuses coarse- and fine-grained representations of entity knowledge from Knowledge Graphs (KGs). Firstly, we identify entities in news content and link them to entities in KGs. Then, a subgraph of KGs is extracted to provide structured knowledge of entities in KGs and fed into a graph neural network to obtain the coarse-grained knowledge representation. This subgraph is pruned to provide fine-grained knowledge and fed into the attentive graph and graph pooling layer. Finally, we integrate the coarse- and fine-grained entity knowledge representations with the textual representation for fake news detection. The experimental results on two benchmark datasets show that our method is superior to state-of-the-art baselines. In addition, it is competitive in the few-shot scenario.

In law, lore, and everyday life, loopholes are commonplace. When people exploit a loophole, they understand the intended meaning or goal of another person, but choose to go with a different interpretation. Past and current AI research has shown that artificial intelligence engages in what seems superficially like the exploitation of loopholes, but this is likely anthropomorphization. It remains unclear to what extent current models, especially Large Language Models (LLMs), capture the pragmatic understanding required for engaging in loopholes. We examined the performance of LLMs on two metrics developed for studying loophole behavior in humans: evaluation (ratings of trouble, upset, and humor), and generation (coming up with new loopholes in a given context). We conducted a fine-grained comparison of state-of-the-art LLMs to humans, and find that while many of the models rate loophole behaviors as resulting in less trouble and upset than outright non-compliance (in line with adults), they struggle to recognize the humor in the creative exploitation of loopholes in the way that humans do. Furthermore, only two of the models, GPT 3 and 3.5, are capable of generating loopholes of their own, with GPT3.5 performing closest to the human baseline.

Designing systems that can reason across cultures requires that they are grounded in the norms of the contexts in which they operate. However, current research on developing computational models of social norms has primarily focused on American society. Here, we propose a novel approach to discover and compare descriptive social norms across Chinese and American cultures. We demonstrate our approach by leveraging discussions on a Chinese Q\&A platform—知乎 (Zhihu)—and the existing SocialChemistry dataset as proxies for contrasting cultural axes, align social situations cross-culturally, and extract social norms from texts using in-context learning. Embedding Chain-of-Thought prompting in a human-AI collaborative framework, we build a high-quality dataset of 3,069 social norms aligned with social situations across Chinese and American cultures alongside corresponding free-text explanations. To test the ability of models to reason about social norms across cultures, we introduce the task of explainable social norm entailment, showing that existing models under 3B parameters have significant room for improvement in both automatic and human evaluation. Further analysis of cross-cultural norm differences based on our dataset shows empirical alignment with the social orientations framework, revealing several situational and descriptive nuances in norms across these cultures.

With the evolution of Large Language Models (LLMs) we can solve increasingly more complex NLP tasks across various domains, including spreadsheets. This work investigates whether LLMs can generate code (Excel OfficeScripts, a TypeScript API for executing many tasks in Excel) that solves Excel specific tasks provided via natural language user instructions. To do so we introduce a new large-scale benchmark, InstructExcel, created by leveraging the ‘Automate’ feature in Excel to automatically generate OfficeScripts from users’ actions. Our benchmark includes over 10k samples covering 170+ Excel operations across 2,000 publicly available Excel spreadsheets. Experiments across various zero-shot and few-shot settings show that InstructExcel is a hard benchmark for state of the art models like GPT-4. We observe that (1) using GPT-4 over GPT-3.5, (2) providing more in-context examples, and (3) dynamic prompting can help improve performance on this benchmark.

Large language models (LLMs) have demonstrated exceptional performance in reasoning tasks with dedicated Chain-of-Thought (CoT) prompts. Further enhancing CoT prompts with exquisite exemplars can significantly improve reasoning performance.However, the effectiveness of CoT prompts may fluctuate dramatically with different choices of in-context examples. Additionally, manual construction of rationale steps can be time-consuming, presenting challenges for the widespread adoption of CoT prompting. In this work, we propose a novel approach by introducing information entropy (IE) as a criteria on for CoT prompt selection. We extend this criterion to the CoT generation and inference stages, automatically generating CoT prompts with higher information entropy scores and adaptively determining the number of samples. These three stages together form our proposed information- entropy-based multi-step reasoning for large language models, named INFORM. Our experiments across seven reasoning benchmarks utilizing two language models(GPT-3.5-Turbo and text-davinci-003) demonstrate the superiority of INFORM both in performance and efficiency.

Large language models have demonstrated exceptional language understanding capabilities in many NLP tasks. Sparsely activated mixture-of-experts (MoE) has emerged as a promising solution for scaling models while maintaining a constant number of computational operations. Existing MoE models adopt a fixed gating network where each token is computed by the same number of experts. This contradicts our intuition that the tokens in each sequence vary in terms of their linguistic complexity and, consequently, require different computational costs. Little is discussed in prior research on the trade-off between computation per token and model performance. This paper introduces adaptive gating in MoE, a flexible training strategy that allows tokens to be processed by a variable number of experts based on expert probability distribution. Adaptive gating preserves sparsity while improving training efficiency. We further draw upon curriculum learning to better align the order of training samples and maximize the training time savings. Extensive experiments on diverse NLP tasks show that adaptive gating reduces at most 22.5% training time while maintaining inference quality. Moreover, we conduct a comprehensive analysis of the gating decisions and present our insights on which tokens are inherently difficult to process, depending on the specific language task.

With recent advances in large language models (LLMs), the concept of automatically generating children’s educational materials has become increasingly realistic. Working toward the goal of age-appropriate simplicity in generated educational texts, we first examine the ability of several popular LLMs to generate stories with properly adjusted lexical and readability levels. We find that, in spite of the growing capabilities of LLMs, they do not yet possess the ability to limit their vocabulary to levels appropriate for younger age groups. As a second experiment, we explore the ability of state-of-the-art lexical simplification models to generalize to the domain of children’s stories and, thus, create an efficient pipeline for their automatic generation. In order to test these models, we develop a dataset of child-directed lexical simplification instances, with examples taken from the LLM-generated stories in our first experiment. We find that, while the strongest-performing current lexical simplification models do not perform as well on material designed for children due to their reliance on large language models behind the scenes, some models that still achieve fairly strong results on general data can mimic or even improve their performance on children-directed data with proper fine-tuning, which we conduct using our newly created child-directed simplification dataset.