Austria

Large Language Models (LLMs) have made notable progress in mathematical reasoning, yet they often rely on single-paradigm reasoning that limits their effectiveness across diverse tasks. In this paper, we introduce Chain-of-Reasoning (CoR), a novel unified framework that integrates multiple reasoning paradigms --- Natural Language Reasoning (NLR), Algorithmic Reasoning (AR), and Symbolic Reasoning (SR) --- to enable synergistic collaboration. CoR generates multiple potential answers using different reasoning paradigms and synthesizes them into a coherent final solution. We propose a Progressive Paradigm Training (PPT) strategy that allows models to progressively master these paradigms, culminating in the development of at CoR-Math-7B. Experimental results demonstrate that CoR-Math-7B significantly outperforms current SOTA models, achieving up to a 41.0% absolute improvement over GPT-4o in theorem proving tasks and a 15% improvement over RL-based methods on the MATH benchmark in arithmetic tasks. These results show the enhanced mathematical comprehensive ability of our model, enabling zero-shot generalization across tasks.The code is available at https://github.com/microsoft/CoR.

ACL 2025

Chain-of-Reasoning: Towards Unified Mathematical Reasoning in Large Language Models via a Multi-Paradigm Perspective

math qa

logical reasoning

reasoning

generalization

poster

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2025". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you need to log in with the **email address you registered with**. 

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

LMs’ alignment with human reading behavior (i.e. psychometric predictive power; PPP) is known to improve during pretraining up to a tipping point, beyond which it either plateaus or degrades. Various factors, such as word frequency, recency bias in attention, and context size, have been theorized to affect PPP, yet there is no current account that explains why such a tipping point exists, and how it interacts with LMs’ pretraining dynamics more generally. We hypothesize that the underlying factor is a pretraining phase transition, characterized by the rapid emergence of specialized attention heads. We conduct a series of correlational and causal experiments to show that such a phase transition is responsible for the tipping point in PPP. We then show that, rather than producing attention patterns that contribute to the degradation in PPP, phase transitions alter the subsequent learning dynamics of the model, such that further training keeps damaging PPP.

Language Models Grow Less Humanlike beyond Phase Transition

Disinformation detection is a key aspect of media literacy. Psychological studies have shown that knowledge of persuasive fallacies helps individuals detect disinformation. Inspired by these findings, we experimented with large language models (LLMs) to test whether infusing persuasion knowledge enhances disinformation detection. As a result, we introduce the Persuasion-Augmented Chain of Thought (PCoT), a novel approach that leverages persuasion to improve disinformation detection in zero-shot classification. We extensively evaluate PCoT on online news and social media posts. Moreover, we publish two novel, up-to-date disinformation datasets: EUDisinfo and MultiDis. These datasets enable the evaluation of PCoT on content entirely unseen by the LLMs used in our experiments, as the content was published after the models' knowledge cutoffs. We show that, on average, PCoT outperforms competitive methods by 15\% across five LLMs and five datasets. These findings highlight the value of persuasion in strengthening zero-shot disinformation detection.

PCoT: Persuasion-Augmented Chain of Thought for Detecting Fake News and Social Media Disinformation

Gestures are an integral part of non-verbal communication, with meanings that vary across cultures, and misinterpretations that can have serious social and diplomatic consequences. As AI systems become more integrated into global applications, ensuring they do not inadvertently perpetuate cultural offenses is critical. To this end, we introduce Multi-Cultural Set of Inappropriate Gestures and Nonverbal Signs (MC-SIGNS), a dataset of 288 gesture-country pairs annotated for offensiveness, cultural significance, and contextual factors across 25 gestures and 85 countries. Through systematic evaluation using MC-SIGNS, we uncover critical limitations: text-to-image (T2I) systems exhibit strong US-centric biases, performing better at detecting offensive gestures in US contexts than in non-US ones; large language models (LLMs) tend to over-flag gestures as offensive; and vision-language models (VLMs) default to US-based interpretations when responding to universal concepts like wishing someone luck, frequently suggesting culturally inappropriate gestures. These findings highlight the urgent need for culturally-aware AI safety mechanisms to ensure equitable global deployment of AI technologies.

Mind the Gesture: Evaluating AI Sensitivity to Culturally Offensive Non-Verbal Gestures

Podcasts provide highly diverse content to a massive listener base through a unique on-demand modality. However, limited data has prevented large-scale computational analysis of the podcast ecosystem. To fill this gap, we introduce a massive dataset of over 1.1M podcast transcripts that is largely comprehensive of all English language podcasts available through public RSS feeds from May and June of 2020. This data is not limited to text, but includes metadata, inferred speaker roles, and audio features and speaker turns for a subset of 370K episodes. Using this data, we conduct a foundational investigation into the content, structure, and responsiveness of this ecosystem. Together, our data and analyses open the door to continued computational research of this popular and impactful medium.

Mapping the Podcast Ecosystem with the Structured Podcast Research Corpus

Natural Language Processing (NLP) is an established and dynamic field. Despite this, what constitutes NLP research remains debated. 
In this work, we address the question by quantitatively examining NLP research papers. We propose a taxonomy of research contributions and introduce _NLPContributions_, a dataset of nearly $2k$ NLP research paper abstracts, carefully annotated to identify scientific contributions and classify their types according to this taxonomy. We also introduce a novel task of automatically identifying contribution statements and classifying their types from research papers. We present experimental results for this task and apply our model to $\sim$$29k$ NLP research papers to analyze their contributions, aiding in the understanding of the nature of NLP research. We show that NLP research has taken a winding path --- with the focus on language and human-centric studies being prominent in the 1970s and 80s, tapering off in the 1990s and 2000s, and starting to rise again since the late 2010s. Alongside this revival, we observe a steady rise in dataset and methodological contributions since the 1990s, such that today, on average, individual NLP papers contribute in more ways than ever before. Our dataset and analyses offer a powerful lens for tracing research trends and offer potential for generating informed, data-driven literature surveys.

The Nature of NLP: Analyzing Contributions in NLP Papers

We introduce the concept of the $\textit{self-referencing causal cycle}$ (abbreviated $\textit{ReCall}$ )—a mechanism that enables large language models (LLMs) to bypass the limitations of unidirectional causality, which underlies a phenomenon known as the $\textit{reversal curse}$. When an LLM is prompted with sequential data, it often fails to recall preceding context. For example, when we ask an LLM to recall the line preceding "O say does that star-spangled banner yet wave" in the U.S. National Anthem, it often fails to correctly return "Gave proof through the night that our flag was still there"—this is due to the reversal curse. It occurs because language models such as ChatGPT and Llama generate text based on preceding tokens, requiring facts to be learned and reproduced in a consistent token order. While the reversal curse is often viewed as a limitation, we offer evidence of an alternative view: it is not always an obstacle in practice. We find that $\textit{ReCall}$ is driven by what we designate as $\textit{cycle tokens}$—sequences that connect different parts of the training data, enabling recall of preceding tokens from succeeding ones. Through rigorous probabilistic formalization and controlled experiments, we demonstrate how the cycles they induce influence a model's ability to reproduce information. To facilitate reproducibility, we provide our code and experimental details at https://anonymous.4open.science/r/remember-B0B8/.

Library-Like Behavior In Language Models is Enhanced by Self-Referencing Causal Cycles

Jailbreaking in Large Language Models (LLMs) is a major security concern as it can deceive LLMs into generating harmful text. However, understanding of how jailbreaking works remains limited, hindering the development of effective defense strategies. To address this issue, we conduct a large-scale analysis of seven different jailbreak methods and identify that disagreements among methods stem from insufficient observation samples.

We introduce the concept of a safety boundary and discover that jailbreaks shift harmful activations outside this boundary, where LLMs become less sensitive to harmful information. Our analysis reveals that low and middle layers play a critical role in these shifts, while deeper layers have a lesser impact.

Building on these insights, we propose a novel defense mechanism called Activation Boundary Defense (ABD), which adaptively constrains activations within the safety boundary. To enhance its effectiveness, we use Bayesian optimization to selectively apply the defense to the low and middle layers.

Experiments on several benchmark datasets demonstrate that ABD achieves an average Defense Success Rate (DSR) of over 98% against various jailbreak attacks, with less than a 2% impact on the model’s general capabilities.

Shaping the Safety Boundaries: Understanding and Defending Against Jailbreaks in Large Language Models

This work evaluates the potential of large language models (LLMs) to power digital assistants capable of complex action execution. Such assistants rely on pre-trained programming knowledge to execute multi-step goals by composing objects and functions defined in assistant libraries into action execution programs. To achieve this, we develop ASPERA, a framework comprising an assistant library simulation and a human-assisted LLM data generation engine. Our engine allows developers to guide LLM generation of high-quality tasks consisting of complex user queries, simulation state and corresponding validation programs, tackling data availability and evaluation robustness challenges. Alongside the framework we release Asper-Bench, an evaluation dataset of $250$ challenging tasks generated using ASPERA, which we use to show that program generation grounded in custom assistant libraries is a significant challenge to LLMs compared to dependency-free code generation.

ASPERA: A Simulated Environment to Evaluate Planning for Complex Action Execution

Large Language Models (LLMs) have demonstrated impressive performance in biomedical relation extraction, even in zero-shot scenarios. However, evaluating LLMs in this task remains challenging due to their ability to generate human-like text, often producing synonyms or abbreviations of gold-standard answers, making traditional automatic evaluation metrics unreliable. On the other hand, while human evaluation is more reliable, it is costly and time-consuming, making it impractical for real-world applications. This paper investigates the use of LLMs-as-the-Judge as an alternative evaluation method for biomedical relation extraction. We benchmark 8 LLMs as judges to evaluate the responses generated by 5 other LLMs across 3 biomedical relation extraction datasets. Unlike other text-generation tasks, we observe that LLM-based judges perform quite poorly (usually below 50% accuracy) in the biomedical relation extraction task. Our findings reveal that it happens mainly because relations extracted by LLMs do not adhere to any standard format. To address this, we propose structured output formatting for LLM-generated responses that helps LLM-Judges to improve their performance by about 15% (on average). We also introduce a domain adaptation technique to further enhance LLM-Judge performance by effectively transferring knowledge between datasets. We release both our human-annotated and LLM-annotated judgment data (36k samples in total) for public use here: https://github.com/tahmedge/llm_judge_biomedical_re.

Improving Automatic Evaluation of Large Language Models (LLMs) in Biomedical Relation Extraction via LLMs-as-the-Judge

This paper focuses on a new task of answering geographic reasoning questions based on the given image (called GeoVQA). Unlike traditional VQA tasks, GeoVQA asks for details about the image-related culture, landscape, etc. This requires not only the identification of the objects in the image, their properties and relations, but also the understanding of the geographic knowledge of the objects, such as location, transportation, landmark, cuisine, etc. This background knowledge does not explicitly appear in the image, nor is there an extra-textual description. Without this missing but necessary knowledge, it is difficult for existing matching-based methods to infer the correct answer. To tackle these challenges, we propose a new geographic reasoning framework for our task. We first analyze the image and describe its fine-grained content by text and keywords using a multi-modal retrieval augmented technique, so as to deduce an answer in a unified textual modality. Next, we retrieve the crucial geographic commonsense knowledge. To reduce the retrieval complexity, we design a dynamic method that can adaptively collect the relevant clues for each reasoning step. The step in the incorrect direction will be pruned according to some judgment criteria. The remaining steps can help us form a reasoning chain to derive a correct answer. Moreover, we create a large-scale dataset GVQA with 41,329 samples to conduct the evaluation. The results demonstrate the effectiveness of our approach.

Downloads

Next from ACL 2025

Language Models Grow Less Humanlike beyond Phase Transition

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES