ACL 2025 Papers

Legal text poses distinctive challenges for natural language processing. The legal import of a term may depend on omissions, cross-references, or silence, Further, legal text is often susceptible to multiple valid, conflicting interpretations; as the saying goes: a good lawyer’s answer to any question is “it depends.”This work introduces a new, hand-coded dataset for the interpretation of privacy policies. It includes privacy policies from 149 firms, including materials incorporated by reference. The policies are annotated across 64 dimension that reflect the applicable legal rules and contested terms from EU and US privacy regulation and litigation. Our annotation methodology is designed to capture the capture core challenges peculiar to legal language, including indeterminacy, interdependence between clauses, meaningful silence, and the implications of legal defaults. We present a set of baseline results for the dataset using current large language models.

Language is a form of symbolic capital that affects people’s lives in many ways (Bourdieu1977,1991). As a powerful means of communication, it reflects identities, cultures, traditions, and societies more broadly. Therefore, data in a given language should be regarded as more than just a collection of tokens. Rigorous data collection and labeling practices are essential for developing more human-centered and socially aware technologies. Although there has been growing interest in under-resourced languages within the NLP community, work in this area faces unique challenges, such as data scarcity and limited access to qualified annotators.In this paper, we collect feedback from individuals directly involved in and impacted by NLP artefacts for medium- and low-resource languages. We conduct both quantitative and qualitative analyses of their responses and highlight key issues related to: (1) data quality, including linguistic and cultural appropriateness; and (2) the ethics of common annotation practices, such as the misuse of participatory research. Based on these findings, we make several recommendations for creating high-quality language artefacts that reflect the cultural milieu of their speakers, while also respecting the dignity and labor of data workers.

Existing LMs struggle with proof-oriented programming due to data scarcity, which manifest in two key ways: (1) a lack of sufficient corpora for proof-oriented programming languages such as F*, and (2) the absence of large-scale, project-level proof-oriented implementations that can teach the model the intricate reasoning process when performing proof-oriented programming. We present the first on synthetic data augmentation for project level proof oriented programming for both generation and repair. Our method addresses data scarcity by synthesizing basic proof-oriented programming problems for proficiency in that language; incorporating diverse coding data for reasoning capability elicitation and creating new proofs and repair data within existing repositories. This approach enables language models to both synthesize and repair proofs for function- and repository-level code. We show that our fine-tuned 14B parameter model, PoPilot, can exceed the performance of the models that outperforms GPT-4o in project-level proof-oriented programming by 64% relative margin, and can improve GPT-4o’s performance by 54% by repairing its outputs over GPT-4o’s self-repair.

We introduce Browsing Lost Unformed Recollections, a tip-of-the-tongue known-item search and reasoning benchmark for general AI assistants. BLUR introduces a set of 573 real-world validated questions that demand searching and reasoning across multimodal and multilingual inputs, as well as proficient tool use, in order to excel on. Humans easily ace these questions (scoring on average 98%), while the best-performing system scores around 56%. To facilitate progress toward addressing this challenging and aspirational use case for general AI assistants, we release 350 questions through a public leaderboard, retain the answers to 250 of them, and have the rest as a private test set.

Automating web navigation which aims to build a web agent that follows user instructions to complete tasks like booking flights by interacting with websites, has received increasing attention due to its practical value. Although existing web agents are mostly equipped with visual perception, planning, and memory abilities, their reasoning process are still deviate from human cognition. In this work, we study the human thought pattern to empower agent with more human-like abilities in web navigation. To tackle this problem, we propose a novel multimodal web agent framework called WebExperT, which is designed to emulate the human planning process of “thinking fast and slow” to effectively decompose complex user instructions. Furthermore, WebExperT leverages experiential learning by reflecting from failure for continuously refining planning and decision-making outcomes. Experimental results on the Mind2Web benchmark demonstrate the superiority of WebExperT in both supervised and unsupervised settings.

A core part of legal work that has been underexplored in Legal NLP is the writing and editing of legal briefs. This requires not only a thorough understanding of the law of a jurisdiction, from judgments to statutes, but also the ability to make new arguments to try to expand the law in a new direction and make novel and creative arguments that are persuasive to judges. To capture and evaluate these legal skills in language models, we introduce BRIEFME, a new dataset focused on legal briefs. It contains three tasks for language models to assist legal professionals in writing briefs: argument summarization, argument completion, and case retrieval. In this work, we describe the creation of these tasks, analyze them, and show how current models perform. We see that today’s large language models (LLMs) are already quite good at the summarization and guided completion tasks, even beating human-generated headings. Yet, they perform poorly on other tasks in our benchmark: realistic argument completion and retrieving relevant legal cases. We hope this dataset encourages more development in Legal NLP in ways that will specifically aid people in performing legal work.

Large Language Models (LLMs) have shown remarkable capabilities in natural language processing but exhibit significant performance gaps among different languages. Most existing approaches to address these disparities rely on pretraining or fine-tuning, which are resource-intensive. To overcome these limitations without incurring significant costs, we propose Inference-Time Cross-Lingual Intervention (INCLINE), a novel framework that enhances LLM performance on low-performing (source) languages by aligning their internal representations with those of high-performing (target) languages during inference. INCLINE initially learns alignment matrices using parallel sentences from source and target languages through a Least-Squares optimization, and then applies these matrices during inference to transform the low-performing language representations toward the high-performing language space. Extensive experiments on nine benchmarks with five LLMs demonstrate that INCLINE significantly improves performance across diverse tasks and languages, compared to recent strong baselines. Our analysis demonstrates that INCLINE is highly cost-effective and applicable to a wide range of applications. In addition, we release the code to foster research along this line.

Despite advancements in Natural Language Processing (NLP), models remain vulnerable to adversarial attacks, such as synonym substitutions. While prior work has focused on improving robustness for feed-forward and convolutional architectures, the robustness of recurrent networks and modern state space models (SSMs), such as S4, remains understudied. These architectures pose unique challenges due to their sequential processing and complex parameter dynamics. In this paper, we introduce a novel regularization technique based on Growth Bound Matrices (GBM) to improve NLP model robustness by reducing the impact of input perturbations on model outputs. We focus on computing the GBM for three architectures: Long Short-Term Memory (LSTM), State Space models (S4), and Convolutional Neural Networks (CNN). Our method aims to (1) enhance resilience against word substitution attacks, (2) improve generalization on clean text, and (3) providing the first systematic analysis of SSM (S4) robustness. Extensive experiments across multiple architectures and benchmark datasets demonstrate that our method improves adversarial robustness by up to (8.8%) over existing baselines. These results highlight the effectiveness of our approach, outperforming several state-of-the-art methods in adversarial defense. Codes are available at https://github.com/BouriMohammed/GBM

The reranker and generator are two critical components in the Retrieval-Augmented Generation (i.e., RAG) pipeline, responsible for ranking relevant documents and generating responses. However, due to differences in pre-training data and objectives, there is an inevitable gap between the documents ranked as relevant by the reranker and those required by the generator to support answering the query. To address this gap, we propose RADIO, a novel and practical preference alignment framework with RAtionale DIstillatiOn. Specifically, We first propose a rationale extraction method that leverages the reasoning capabilities of large language models (LLMs) to extract the rationales necessary for answering the query. Subsequently, a rationale-based alignment process is designed to rerank the documents based on the extracted rationales, and fine-tune the reranker to align the preferences. We conduct extensive experiments on two tasks across three datasets to demonstrate the effectiveness of our approach compared to baseline methods. Our code is released online to ease reproduction.

Large language models (LLMs)-based personal assistants may struggle to effectively utilize long-term conversational histories.Despite advances in long-term memory systems and dense retrieval methods, these assistants still fail to capture entity relationships and handle multiple intents effectively. To tackle above limitations, we propose **Associa**, a graph-structured memory framework that mimics human cognitive processes. Associa comprises an event-centric memory graph and two collaborative components: **Intuitive Association**, which extracts evidence-rich subgraphs through Prize-Collecting Steiner Tree optimization, and **Deliberating Recall**, which iteratively refines queries for comprehensive evidence collection. Experiments show that Associa significantly outperforms existing methods in retrieval and QA (question and answering) tasks across long-term dialogue benchmarks, advancing the development of more human-like AI memory systems.

Most LLMs universally excel at generating code for high-resource programming languages (HRPLs) like Python, a capability that has become standard due to the abundance of training data. However, they struggle significantly with low-resource programming languages (LRPLs) such as D, exacerbating the digital divide. This gap limits developers using LRPLs from equally benefiting and hinders innovation within underrepresented programming communities. To make matters worse, manually generating data for LRPLs is highly labor intensive and requires expensive expert effort. In this work, we begin by analyzing the NL-PL Gap, where LLMs’ direct-generated LRPL data often suffers from subpar quality due to the misalignment between natural language (NL) instructions and programming language (PL) outputs. To address this issue, we introduce Bridge-Assist Generation, a method to generate LRPL data utilizing LLM’s general knowledge, HRPL proficiency, and in-context learning capabilities. To further maximize the utility of the generated data, we propose Bridged Alignment to obtain Bridge-Coder. To thoroughly evaluate our approach, we select four relatively LRPLs: R, D, Racket, and Bash. Experimental results reveal that Bridge-Coder achieves significant improvements over the original model, with average gains of 18.71 and 10.81 on two comprehensive benchmarks, M-HumanEval and M-MBPP.

Recent Large Language Models (LLMs) have demonstrated impressive translation performance without requiring fine-tuning on additional parallel corpora. However, they still face significant challenges in certain scenarios, particularly when translating low-resource languages. A common approach to address this issue is to provide external knowledge, such as few-shot examples, to assist LLMs in translating specific source sentences. However, this method is fundamentally limited by the quality or quantity of relevant sources, which cannot always be guaranteed. To reduce LLMs’ reliance on external sources, we propose BridG MT, a method that combines Sentence Bridging, which generates a sequence of sentences as a bridge that gradually transition from easy-to-translate to more difficult, and Gradual MT, which sequentially translates these sentences using earlier translations as few-shot examples for subsequent ones. Experiments conducted on four LLMs across seven languages demonstrate that our method effectively enhances translation performance, even outperforming translation methods that rely on a large number of few-shot examples.

A significant portion of the energy consumed by Large Language Models (LLMs) arises from their inference processes; hence developing energy-efficient methods for inference is crucial. While several techniques exist for inference optimization, output compression remains relatively unexplored, with only a few preliminary efforts addressing this aspect. In this work, we first benchmark 12 decoder-only LLMs across 5 datasets, revealing that these models often produce responses that are substantially longer than necessary. We then conduct a comprehensive quality assessment of LLM responses, formally defining six information categories present in LLM responses. We show that LLMs often tend to include redundant or additional information besides the minimal answer. To address this issue of long responses by LLMs, we explore several simple and intuitive prompt-engineering strategies.Empirical evaluation shows that appropriate prompts targeting length reduction and controlling information content can achieve significant energy optimization between 25-60% by reducing the response length while preserving the quality of LLM responses.

We propose a novel discriminative model for sequence labeling called Bregman conditional random fields (BCRF).Contrary to standard linear-chain conditional random fields,BCRF allows fast parallelizable inference algorithms based on iterative Bregman projections.We show how such models can be learned using Fenchel-Young losses, including extension for learning from partial labels.Experimentally, our approach delivers comparable results to CRF while being faster, and achieves better results in highly constrained settings compared to mean field, another parallelizable alternative.

The release of OpenAI’s O1 and subsequent projects like DeepSeek R1 has significantly advanced research on complex reasoning in LLMs. This paper systematically analyzes existing reasoning studies from the perspective of self-evolution, structured into three components: data evolution, model evolution, and self-evolution. Data evolution explores methods to generate higher-quality reasoning training data. Model evolution focuses on training strategies to boost reasoning capabilities. Self-evolution research autonomous system evolution via iterating cycles of data and model evolution. We further discuss the scaling law of self-evolution and analyze representative O1-like works through this lens. By summarizing advanced methods and outlining future directions, this paper aims to drive advancements in LLMs’ reasoning abilities.

Large Language Models (LLMs), despite advanced general capabilities, still suffer from numerous safety risks, especially jailbreak attacks that bypass safety protocols. Understanding these vulnerabilities through black-box jailbreak attacks, which better reflect real-world scenarios, offers critical insights into model robustness. While existing methods have shown improvements through various prompt engineering techniques, their success remains limited against safety-aligned models, overlooking a more fundamental problem: the effectiveness is inherently bounded by the predefined strategy spaces. However, expanding this space presents significant challenges in both systematically capturing essential attack patterns and efficiently navigating the increased complexity. To better explore the potential of expanding the strategy space, we address these challenges through a novel framework that decomposes jailbreak strategies into essential components based on the Elaboration Likelihood Model (ELM) theory and develops genetic-based optimization with intention evaluation mechanisms. To be striking, our experiments reveal unprecedented jailbreak capabilities by expanding the strategy space: we achieve over 90% success rate on Claude-3.5 where prior methods completely fail, while demonstrating strong cross-model transferability and surpassing specialized safeguard models in evaluation accuracy. The code is open-sourced at: https://github.com/Aries-iai/CL-GSO.

Current EEG/MEG-to-text decoding systems suffer from three key limitations: (1) reliance on teacher-forcing methods, which compromises robustness during inference, (2) sensitivity to session-specific noise, hindering generalization across subjects, and (3) misalignment between brain signals and linguistic representations due to pre-trained language model over-dominance. To overcome these challenges, we propose BrainECHO (Brain signal decoding via vEctor-quantized speCtrogram reconstruction for WHisper-enhanced text generatiOn), a multi-stage framework that employs decoupled representation learning to achieve state-of-the-art performance on both EEG and MEG datasets. Specifically, BrainECHO consists of three stages: (1) Discrete autoencoding, which transforms continuous Mel spectrograms into a finite set of high-quality discrete representations for subsequent stages. (2) Frozen alignment, where brain signal embeddings are mapped to corresponding Mel spectrogram embeddings in a frozen latent space, effectively filtering session-specific noise through vector-quantized reconstruction, yielding a 3.65% improvement in BLEU-4 score. (3) Constrained decoding fine-tuning, which leverages the pre-trained Whisper model for audio-to-text translation, balancing signal adaptation with knowledge preservation, and achieving 74%-89% decoding BLEU scores without excessive reliance on teacher forcing. BrainECHO demonstrates robustness across sentence, session, and subject-independent conditions, passing Gaussian noise tests and showcasing its potential for enhancing language-based brain-computer interfaces.

Humor is prevalent in online communications and it often relies on more than one modality (e.g., cartoons and memes).Interpreting humor in multimodal settings requires drawing on diverse types of knowledge, including metaphorical, sociocultural, and commonsense knowledge. However, identifying the most useful knowledge remains an open question. We introduce BottleHumor, a method inspired by the information bottleneck principle that elicits relevant world knowledge from vision and language models which is iteratively refined for generating an explanation of the humor in an unsupervised manner. Our experiments on three datasets confirm the advantage of our method over a range of baselines. Our method can further be adapted in the future for additional tasks that can benefit from eliciting and conditioning on relevant world knowledge and open new research avenues in this direction.

Large language models (LLMs) demonstrate considerable proficiency in numerous coding-related tasks; however, their capabilities in detecting software vulnerabilities remain limited. This limitation primarily stems from two factors: (1) the absence of reasoning data related to vulnerabilities, which hinders the models’ ability to capture underlying vulnerability patterns; and (2) their focus on learning semantic representations rather than the reason behind them, thus failing to recognize semantically similar vulnerability samples. Furthermore, the development of LLMs specialized in vulnerability detection is challenging, particularly in environments characterized by the scarcity of high-quality datasets. In this paper, we propose a novel framework ReVD that excels at mining vulnerability patterns through reasoning data synthesizing and vulnerability-specific preference optimization. Specifically, we construct forward and backward reasoning processes for vulnerability and corresponding fixed code, ensuring the synthesis of high-quality reasoning data. Moreover, we design the triplet supervised fine-tuning followed by curriculum online preference optimization for enabling ReVD to better understand vulnerability patterns. The extensive experiments conducted on PrimeVul and SVEN datasets demonstrate that ReVD sets new state-of-the-art for LLM-based software vulnerability detection, e.g., 12.24%-22.77% improvement in the accuracy. The source code and data are available at https://github.com/Xin-Cheng-Wen/PO4Vul.