COLING 2025 论文列表

In medical question-answering, traditional knowledge triples often fail due to superfluous data and their inability to capture complex relationships between symptoms and treatments across diseases. This limits models’ ability to provide accurate, contextually relevant responses. To overcome this, we introduce MedEx, which employs First-Order Logic (FOL)-based reasoning to model intricate relationships between diseases and treatments. We construct FOL-based triplets that encode the interplay of symptoms, diseases, and treatments, capturing not only surface-level data but also the logical constraints of the medical domain. MedEx encodes the discourse (questions and context) using a transformer-based unit, enhancing context comprehension. These encodings are processed by a Knowledge Injection Cell that integrates knowledge graph triples via a Graph Attention Network. The Logic Fusion Cell then combines medical-specific logical rule triples (e.g., co-occurrence, causation, diagnosis) with knowledge triples and extracts answers through a feed-forward layer. Our analysis demonstrates MedEx’s effectiveness and generalization across medical question-answering tasks. By merging logical reasoning with knowledge, MedEx provides precise medical answers and adapts its logical rules based on training data nuances.

Advancements in dialogue systems powered by large language models (LLMs) have outpaced the development of reliable evaluation metrics, particularly for diverse and creative responses. We present a benchmark for evaluating the robustness of reference-free dialogue metrics against four categories of adversarial attacks: speaker tag prefixes, static responses, ungrammatical responses, and repeated conversational context. We analyze metrics such as DialogRPT, UniEval, and PromptEval—a prompt-based method leveraging LLMs—across grounded and ungrounded datasets. By examining both their correlation with human judgment and susceptibility to adversarial attacks, we find that these two axes are not always aligned; metrics that appear to be equivalent when judged by traditional benchmarks may, in fact, vary in their scores of adversarial responses. These findings motivate the development of nuanced evaluation frameworks to address real-world dialogue challenges.

To address an important gap in creating children’s stories for vocabulary enrichment, we investigate the automatic evaluation of how well stories convey the semantics of target vocabulary words, a task with substantial implications for generating educational content. We motivate this task, which we call measuring contextual informativeness in children’s stories, and provide a formal task definition as well as a dataset for the task. We further propose a method for automating the task using a large language model (LLM). Our experiments show that our approach reaches a Spearman correlation of 0.4983 with human judgments of informativeness, while the strongest baseline only obtains a correlation of 0.3534. An additional analysis shows that the LLM-based approach is able to generalize to measuring contextual informativeness in adult-directed text, on which it also outperforms all baselines.

Entity matching (EM) is a critical step in entity resolution (ER). Recently, entity matching based on large language models (LLMs) has shown great promise. However, current LLM-based entity matching approaches typically follow a binary matching paradigm that ignores the global consistency among record relationships. In this paper, we investigate various methodologies for LLM-based entity matching that incorporate record interactions from different perspectives. Specifically, we comprehensively compare three representative strategies: matching, comparing, and selecting, and analyze their respective advantages and challenges in diverse scenarios. Based on our findings, we further design a compound entity matching framework (ComEM) that leverages the composition of multiple strategies and LLMs. ComEM benefits from the advantages of different sides and achieves improvements in both effectiveness and efficiency. Experimental results on 8 ER datasets and 10 LLMs verify the superiority of incorporating record interactions through the selecting strategy, as well as the further cost-effectiveness brought by ComEM.

Considerable advancements have been made to tackle the misrepresentation of information derived from reference articles in the domains of fact-checking and faithful summarization. However, an unaddressed aspect remains - the identification of social media posts that manipulate information within associated news articles. This task presents a significant challenge, primarily due to the prevalence of personal opinions in such posts. We present a novel task, identifying manipulation of news on social media, which aims to detect manipulation in social media posts and identify manipulated or inserted information. To study this task, we have proposed a data collection schema and curated a dataset called ManiTweet, consisting of 3.6K pairs of tweets and corresponding articles. Our analysis demonstrates that this task is highly challenging, with large language models (LLMs) yielding unsatisfactory performance. Additionally, we have developed a simple yet effective basic model that outperforms LLMs significantly on the ManiTweet dataset. Finally, we have conducted an exploratory analysis of human-written tweets, unveiling intriguing connections between manipulation and the domain and factuality of news articles, as well as revealing that manipulated sentences are more likely to encapsulate the main story or consequences of a news outlet.

Indirect User Requests (IURs), such as “It’s cold in here” instead of “Could you please increase the temperature?” are common in human-human task-oriented dialogue and require world knowledge and pragmatic reasoning from the listener. While large language models (LLMs) can handle these requests effectively, smaller models deployed on virtual assistants often struggle due to resource constraints. Moreover, existing task-oriented dialogue benchmarks lack sufficient examples of complex discourse phenomena such as indirectness. To address this, we propose a set of linguistic criteria along with an LLM-based pipeline for generating realistic IURs to test natural language understanding (NLU) and dialogue state tracking (DST) models before deployment in a new domain. We also release IndirectRequests, a dataset of IURs based on the Schema-Guided Dialogue (SGD) corpus, as a comparative testbed for evaluating the performance of smaller models in handling indirect requests.

World models, which encapsulate the dynamics of how actions affect environments, are foundational to the functioning of intelligent agents. In this work, we explore the potential of Large Language Models (LLMs) to operate as world models. Although LLMs are not inherently designed to model real-world dynamics, we show that they can be induced to perform two critical world model functions: determining the applicability of an action based on a given world state, and predicting the resulting world state upon action execution. This is achieved by fine-tuning two separate LLMs—one for precondition prediction and another for effect prediction—while leveraging synthetic data generation techniques. Through human-participant studies, we validate that the precondition and effect knowledge generated by our models aligns with human understanding of world dynamics. We also analyze the extent to which the world model trained on our synthetic data results in an inferred state space that supports the creation of action chains, a necessary property for planning.

The open-domain text-to-SQL task aims to retrieve question-relevant tables from massive databases and generate SQL. However, the performance of current methods is constrained by single-hop retrieval, and existing multi-hop retrieval of open-domain question answering is not directly applicable due to the tendency to retrieve tables similar to the retrieved ones but irrelevant to the question. Since the questions in text-to-SQL usually contain all required information, while previous multi-hop retrieval supplements the questions with retrieved documents. Therefore, we propose the multi-hop table retrieval with removal (MURRE), which removes previously retrieved information from the question to guide the retriever towards unretrieved relevant tables. Our experiments on two open-domain text-to-SQL datasets demonstrate an average improvement of 5.7% over the previous state-of-the-art results.

Graph-enhanced large language models (LLMs) leverage LLMs’ remarkable ability to model language and use graph structures to capture topological relationships. Existing graph-enhanced LLMs typically retrieve similar subgraphs to augment LLMs, where the subgraphs carry the entities related to our target and relations among the entities. However, the retrieving methods mainly focus solely on accurately matching subgraphs between our target subgraph and the candidate subgraphs at the same scale, neglecting that the subgraphs with different scales may also share similar semantics or structures. To tackle this challenge, we introduce a graph-enhanced LLM with multi-scale retrieval (MSG-LLM). It captures similar graph structures and semantics across graphs at different scales and bridges the graph alignment across multiple scales. The larger scales maintain the graph’s global information, while the smaller scales preserve the details of fine-grained sub-structures. Specifically, we construct a multi-scale variation to dynamically shrink the scale of graphs. Further, we employ a graph kernel search to discover subgraphs from the entire graph, which essentially achieves multi-scale graph retrieval in Hilbert space. Additionally, we propose to conduct multi-scale interactions (message passing) over graphs at various scales to integrate key information. The interaction also bridges the graph and LLMs, helping with graph retrieval and LLM generation. Finally, we employ a Chain-of-Thought-based LLM prediction to perform the downstream tasks. We evaluate our approach on two graph-based downstream tasks and the experimental results show that our method achieves state-of-the-art performance.

The complexities of chats, such as the stylized contents specific to source segments and dialogue consistency, pose significant challenges for machine translation. Recognizing the need for a precise evaluation metric to address the issues associated with chat translation, this study introduces Multidimensional Quality Metrics for Chat Translation (MQM-Chat), which encompasses seven error types, including three specifically designed for chat translations: ambiguity and disambiguation, buzzword or loanword issues, and dialogue inconsistency. In this study, human annotations were applied to the translations of chat data generated by five translation models. Based on the error distribution of MQM-Chat and the performance of relabeling errors into chat-specific types, we concluded that MQM-Chat effectively classified the errors while highlighting chat-specific issues explicitly. The results demonstrate that MQM-Chat can qualify both the lexical accuracy and semantical accuracy of translation models in chat translation tasks.

Large Language Models (LLMs) have shown significant potential as judges for Machine Translation (MT) quality assessment, providing both scores and fine-grained feedback. Although approaches such as GEMBA-MQM have shown state-of-the-art performance on reference-free evaluation, the predicted errors do not align well with those annotated by human, limiting their interpretability as feedback signals. To enhance the quality of error annotations predicted by LLM evaluators, we introduce a universal and training-free framework, **MQM-APE**, based on the idea of filtering out non-impactful errors by Automatically Post-Editing (APE) the original translation based on each error, leaving only those errors that contribute to quality improvement. Specifically, we prompt the LLM to act as 1) *evaluator* to provide error annotations, 2) *post-editor* to determine whether errors impact quality improvement and 3) *pairwise quality verifier* as the error filter. Experiments show that our approach consistently improves both the reliability and quality of error spans against GEMBA-MQM, across eight LLMs in both high- and low-resource languages. Orthogonal to trained approaches, MQM-APE complements translation-specific evaluators such as Tower, highlighting its broad applicability. Further analysis confirms the effectiveness of each module and offers valuable insights into evaluator design and LLMs selection.

Large Language Models (LLMs) have demonstrated significant capabilities across numerous application domains. A key challenge is to keep these models updated with latest available information, which limits the true potential of these models for the end-applications. Although, there have been numerous attempts for LLMs’ Knowledge Editing (KE), i.e., to update and/or edit the LLMs’ prior knowledge and in turn test it via Multi-hop Question Answering (MQA), yet so far these studies are primarily focused and/or developed for English language. To bridge this gap, in this paper we propose: Multi-hop Questioning Answering under Knowledge Editing for Arabic Language (MQA-KEAL). MQA-KEAL stores knowledge edits as structured knowledge units in the external memory. In order to solve multi-hop question, it first uses task-decomposition to decompose the question into smaller sub-problems. Later for each sub-problem, it iteratively queries the external memory and/or target LLM in order to generate the final response. In addition, we also contribute MQUAKE-AR (Arabic translation of English benchmark MQUAKE), as well as a new benchmark MQA-AEVAL for rigorous performance evaluation of MQA under KE for Arabic language. Experimentation evaluation reveals MQA-KEAL outperforms the baseline models by a significant margin. We release the codes for MQA-KEAL at https: //github.com/asif6827/MQA-Keal.

Aligning Large Language Models (LLMs) with human feedback is crucial for their development. Existing preference optimization methods such as DPO and KTO, while improved based on Reinforcement Learning from Human Feedback (RLHF), are inherently derived from PPO, requiring a reference model that adds GPU memory resources and relies heavily on abundant preference data. Meanwhile, current preference optimization research mainly targets single-question scenarios with two replies, neglecting optimization with multiple replies, which leads to a waste of data in the application. This study introduces the MPPO algorithm, which leverages the average likelihood of model responses to fit the reward function and maximizes the utilization of preference data. Through a comparison of Point-wise, Pair-wise, and List-wise implementations, we found that the Pair-wise approach achieves the best performance, significantly enhancing the quality of model responses. Experimental results demonstrate MPPO’s outstanding performance across various benchmarks. On MT-Bench, MPPO outperforms DPO, ORPO, and SimPO. Notably, on Arena-Hard, MPPO surpasses DPO and ORPO by substantial margins. These achievements underscore the remarkable advantages of MPPO in preference optimization tasks.

The advancement of social media has intensified interest in the research direction of Multimodal Sentiment Analysis (MSA). However, current methodologies exhibit relative limitations, particularly in their fusion mechanisms that overlook nuanced differences and similarities across modalities, leading to potential biases in MSA. In addition, indiscriminate fusion across modalities can introduce unnecessary complexity and noise, undermining the effectiveness of the analysis. In this essay, a Modal-Preserving and Interaction-Driven Fusion Network is introduced to address the aforementioned challenges. The compressed representations of each modality are initially obtained through a Token Refinement Module. Subsequently, we employ a Dual Perception Fusion Module to integrate text with audio and a separate Adaptive Graded Fusion Module for text and visual data. The final step leverages text representation to enhance composite representation. Our experiments on CMU-MOSI, CMU-MOSEI, and CH-SIMS datasets demonstrate that our model achieves state-of-the-art performance.

How emotions are expressed depends on the context and domain. On X (formerly Twitter), for instance, an author might simply use the hashtag #anger, while in a news headline, emotions are typically written in a more polite, indirect manner. To enable conditional text generation models to create emotionally connotated texts that fit a domain, users need to have access to a parameter that allows them to choose the appropriate way to express an emotion. To achieve this, we introduce MOPO, a Multi-Objective Prompt Optimization methodology. MOPO optimizes prompts according to multiple objectives (which correspond here to the output probabilities assigned by emotion classifiers trained for different domains). In contrast to single objective optimization, MOPO outputs a set of prompts, each with a different weighting of the multiple objectives. Users can then choose the most appropriate prompt for their context. We evaluate MOPO using three objectives, determined by various domain-specific emotion classifiers. MOPO improves performance by up to 15 pp across all objectives with a minimal loss (1–2 pp) for any single objective compared to single-objective optimization. These minor performance losses are offset by a broader generalization across multiple objectives – which is not possible with single-objective optimization. Additionally, MOPO reduces computational requirements by simultaneously optimizing for multiple objectives, eliminating separate optimization procedures for each objective.

Event relation extraction (ERE) is becoming increasingly important in the era of large language models. An extensive body of research has explored how performance can be further enhanced by the emergence of exciting technologies like chain-of-thought and self-refinement. In this paper, we introduce MMD-ERE, a multi-agent multi-sided debate approach for event relation extraction, which explores the understanding of event relations among different participants before and after debate. Specifically, for organizing the debate, participants are divided into multiple groups, each assigned its own debate topic, and the process effectively integrates both cooperation and confrontation. We also regard the audience as a crucial participant, as their conclusions from an observer’s perspective tend to be more objective. In the end, we explore the understanding of event relations among different participants before and after the debate. Experiments across various ERE tasks and LLMs demonstrate that MMD-ERE outperforms established baselines. Further analysis shows that debates can effectively enhance participants’ understanding of event relations.

The extensive utilization of large language models (LLMs) underscores the crucial necessity for precise and contemporary knowledge embedded within their intrinsic parameters. Existing research on knowledge editing primarily concentrates on monolingual scenarios, neglecting the complexities presented by multilingual contexts and multi-hop reasoning. To address these challenges, our study introduces MLaKE (Multilingual Language Knowledge Editing), a novel benchmark comprising 4072 multi-hop and 5360 single-hop questions designed to evaluate the adaptability of knowledge editing methods across five languages: English, Chinese, Japanese, French, and German. MLaKE aggregates fact chains from Wikipedia across languages and utilizes LLMs to generate questions and answer. We assessed the effectiveness of current multilingual knowledge editing methods using the MLaKE dataset. Our results show that due to considerable inconsistencies in both multilingual performance and encoding efficiency, these methods struggle to generalize effectively across languages. The accuracy of these methods when editing English is notably higher than for other languages. The experimental results further demonstrate that models encode knowledge and generation capabilities for different languages using distinct parameters, leading to poor cross-lingual transfer performance in current methods. Transfer performance is notably better within the same language family compared to across different families. These findings emphasize the urgent need to improve multilingual knowledge editing methods.

Combined Image Retrieval (CIR) involves retrieving an image based on a reference image and a brief text description, which is widely present in various scenarios such as fashion recommendation. Existing methods can be mainly divided into two categories, respectively supervised CIR methods and Zero-Shot CIR (ZS-CIR) methods. In contrast to supervised CIR methods, which need manually annotated triples for training task-specific models, ZS-CIR models can be trained using images datasets only and performs well. However, ZS-CIR still faces the primary challenge of learning how to map pseudo-words to images within the joint image-text embedding space. Therefore, in this paper, we propose a novel image-text mapping network, named MLLM-I2W, which adaptively converts description-related image information into pseudo-word markers for precise ZS-CIR. Specifically, the image and text encoding enhancement module within the MLLM prompt selects subject headings and generates text descriptions. It then reduces the modality gap between images and text using uncertainty modeling. An adaptive weighting module and a prototype are proposed to adjust and learn the deep fusion features, which are further mapped to pseudo-word markers via well-designed MOE-based mapping network. Our model demonstrates consistent improvements across common CIR benchmarks, including COCO, CIRR, and Fashion-IQ.

Narrative understanding and story generation are critical challenges in natural language processing (NLP), with much of the existing research focused on summarization and question-answering tasks. While previous studies have explored predicting plot endings and generating extended narratives, they often neglect the logical coherence within stories, leaving a significant gap in the field. To address this, we introduce the Missing Logic Detector by Emotion and Action (MLD-EA) model, which leverages large language models (LLMs) to identify narrative gaps and generate coherent sentences that integrate seamlessly with the story’s emotional and logical flow. The experimental results demonstrate that the MLD-EA model enhances narrative understanding and story generation, highlighting LLMs’ potential as effective logic checkers in story writing with logical coherence and emotional consistency. This work fills a gap in NLP research and advances border goals of creating more sophisticated and reliable story-generation systems.