Austria

Recently, O1-like models have emerged as representative examples, illustrating the effectiveness of long chain-of-thought (CoT) in reasoning tasks such as math and coding tasks. In this paper, we introduce DRT, an attempt to bring the success of long CoT to neural machine translation (MT). Specifically, in view of the literature books that might involve similes and metaphors, translating these texts to a target language is very difficult in practice due to cultural differences. In such cases, literal translation often fails to convey the intended meaning effectively. Even for professional human translators, considerable thought must be given to preserving semantics throughout the translation process. To simulate LLMs&#39; long thought ability in MT, we first mine sentences containing similes or metaphors from existing literature books, and then develop a multi-agent framework to translate these sentences via long thought. In the multi-agent framework, a translator is used to iteratively translate the source sentence under the suggestions provided by an advisor. To ensure the effectiveness of the long thoughts, an evaluator is also employed to quantify the translation quality in each round. In this way, we collect tens of thousands of long-thought MT data, which is used to train our DRT. Using Qwen2.5 and LLama-3.1 as the backbones, DRT models can learn the thought process during machine translation, and outperform vanilla LLMs as well as LLMs which are simply fine-tuning on the paired sentences without long thought, showing its effectiveness.

ACL 2025

DRT: Deep Reasoning Translation via Long Chain-of-Thought

large reasoning models

literary translation

machine translation

### 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)

Recently, O1-like models have emerged as representative examples, illustrating the effectiveness of long chain-of-thought (CoT) in reasoning tasks such as math and coding tasks. In this paper, we introduce DRT, an attempt to bring the success of long CoT to neural machine translation (MT). Specifically, in view of the literature books that might involve similes and metaphors, translating these texts to a target language is very difficult in practice due to cultural differences. In such cases, literal translation often fails to convey the intended meaning effectively. Even for professional human translators, considerable thought must be given to preserving semantics throughout the translation process. To simulate LLMs' long thought ability in MT, we first mine sentences containing similes or metaphors from existing literature books, and then develop a multi-agent framework to translate these sentences via long thought. In the multi-agent framework, a translator is used to iteratively translate the source sentence under the suggestions provided by an advisor. To ensure the effectiveness of the long thoughts, an evaluator is also employed to quantify the translation quality in each round. In this way, we collect tens of thousands of long-thought MT data, which is used to train our DRT. Using Qwen2.5 and LLama-3.1 as the backbones, DRT models can learn the thought process during machine translation, and outperform vanilla LLMs as well as LLMs which are simply fine-tuning on the paired sentences without long thought, showing its effectiveness.

poster

Indigenous languages remain largely invisible in commercial language identification systems—a stark reality exemplified by Google Translate's LangID tool, which supports 109 languages but excludes all 150 Indigenous languages of North America. This technological marginalization is particularly acute for Alaska's 20 Native languages, all of which face endangerment despite their rich linguistic heritage. We present GenAlaskan, a framework demonstrating how both large language models and specialized classifiers can effectively identify these languages with minimal data. Working closely with Native Alaskan community members, we created Akutaq-2k, a carefully curated dataset of 2000 sentences spanning all 20 languages—named after the traditional Yup'ik dessert symbolizing the blending of diverse elements. We design few-shot learning on proprietary and open-source LLMs, achieving nearly 100% accuracy with just 40 examples per language. While initial zero-shot attempts showed limited success, our systematic attention head pruning revealed critical architectural components for accurate language differentiation, providing insights into model decision-making for low-resource languages. Our results challenge the notion that effective Indigenous language identification requires massive resources or corporate infrastructure, demonstrating that targeted technological interventions can drive meaningful progress in preserving endangered languages in the digital age.

Visibility as Survival: Generalizing NLP for Native Alaskan Language Identification

Automatic related work generation (RWG) can save people’s time and effort when writing a draft of related work section (RWS) for further revision. However, existing methods for RWG always suffer from shallow comprehension due to taking the limited portions of references papers as input and isolated explanation for each reference due to ineffective capturing the relationships among them. To address these issues, we focus on full-text-based RWG task and propose a novel multi-agent framework. Our framework consists of three agents: a selector that decides which section of the papers is going to read next, a reader that digests the selected section and updates a shared working memory, and a writer that generates RWS based on the final curated memory. To better capture the relationships among references, we also propose two graph-aware strategies for selector, enabling to optimize the reading order with constrains of the graph structure. Extensive experiments demonstrate that our framework consistently improves performance across three base models and various input configurations. The graph-aware selectors outperform alternative selectors, achieving state-of-the-art results. The code and data are available at https://github.com/1190200817/Full_Text_RWG.

Select, Read, and Write: A Multi-Agent Framework of Full-Text-based Related Work Generation

Translating multi-word expressions (MWEs) and idioms requires a deep understanding of the cultural nuances of both the source and target languages. This challenge is further amplified by the one-to-many nature of idiomatic translations, where a single source idiom can have multiple target-language equivalents depending on cultural references and contextual variations. Traditional static knowledge graphs (KGs) and prompt-based approaches struggle to capture these complex relationships, often leading to suboptimal translations. To address this, we propose an IdiomCE, an adaptive graph neural network (GNN) based methodology that learns intricate mappings between idiomatic expressions, effectively generalizing to both seen and unseen nodes during training. Our proposed method enhances translation quality even in resource-constrained settings, facilitating improved idiomatic translation in smaller models. We evaluate our approach on multiple idiomatic translation datasets using reference-less metrics, demonstrating significant improvements in translating idioms from English to various Indian languages

Graph-Assisted Culturally Adaptable Idiomatic Translation for Indic languages

The generative capabilities of the large language models (LLMs) are deployed for domain-specific question answering systems. However, their ability to answer climate adaptation questions remains unclear. In particular, can they be used by agronomists and climate scientists to answer questions on the best climate adaptation strategies? Answering questions in this domain requires knowledge of climate data and its uncertainties, and the ability to link them to the broader climate literature while accommodating the unique constraints of users and experts. We investigate the generative and evaluative capabilities of several state-of-the-art LLMs, open-source and proprietary, on climate adaptation questions posed by domain experts using evaluation criteria designed by the experts.

We propose an iterative exploration framework that enables LLMs to dynamically aggregate information from heterogeneous sources, such as text from climate literature and structured tabular climate data from climate model projections and historical observations. Our experiments demonstrate that LLMs can aggregate heterogeneous data to (1) answer questions, but at a trade-off between presentation quality and epistemological accuracy; and, (2) evaluate answers, but are not as competent at identifying high-quality answers and erroneous information compared to domain experts.

Question answering in Climate Adaptation: Model Development and Evaluation with Expert Feedback

Autoregressive decoders in large language models (LLMs) excel at capturing users' sequential behaviors for generative recommendations. However, they inherently struggle to leverage graph-structured user-item interactions, which are widely recognized as beneficial. This paper presents AGRec, adapting LLMs' decoders with graph reasoning for recommendation. We reveal that LLMs and graph neural networks (GNNs) manifest complementary strengths in distinct user domains. Building on this, we augment the decoding logits of LLMs with an auxiliary GNN model to optimize token generation. Moreover, we introduce a rankable finite state machine to tackle two challenges: (1) adjusting autoregressive generation with discriminative decoders that directly predict user-item similarity, and (2) token homogeneity, where LLMs often generate items with similar prefix tokens, narrowing the scope of beam search. This approach offers a novel perspective to enhance LLMs with graph knowledge. Our AGRec outperforms state-of-the-art models in sequential recommendations. Our code is available online.

AGRec: Adapting Autoregressive Decoders with Graph Reasoning for LLM-based Sequential Recommendation

The multi-label aspect category detection (MACD) task has attracted great attention in sentiment analysis. Many recent methods have formulated the MACD task by learning robust prototypes to represent categories with limited support samples. However, few of them address the noise categories in the support set that hinder their models from effective prototype generations. To this end, we propose a causal denoising prototypical network (CDPN) for few-shot MACD. We reveal the underlying relation between causal inference and contrastive learning, and present causal contrastive learning (CCL) using discrete and continuous noise as negative samples. We empirically found that CCL can (1) prevent models from overly predicting more categories and (2) mitigate semantic ambiguity issues among categories. Experimental results show that CDPN outperforms competitive baselines. Our code is available online.

Causal Denoising Prototypical Network for Few-Shot Multi-label Aspect Category Detection

With the rapid advancement of Large Language Models (LLMs), there is an increasing need for challenging benchmarks to evaluate their capabilities in handling complex tabular data. However, existing benchmarks are either based on outdated data setups or focus solely on simple, flat table structures. In this paper, we introduce **RealHiTBench**, a comprehensive benchmark designed to evaluate the performance of both LLMs and Multimodal LLMs (MLLMs) across a variety of input formats for complex tabular data, including LaTeX, HTML, and PNG. RealHiTBench also includes a diverse collection of tables with intricate structures, spanning a wide range of task types. Our experimental results, using **20** state-of-the-art LLMs, demonstrate that RealHiTBench is indeed a challenging benchmark. Moreover, we also develop TreeThinker, a tree-based agent that organizes hierarchical headers into a tree structure for enhanced tabular reasoning, validating the importance of improving LLMs' perception of table hierarchies. We hope that our work will inspire further research on tabular data reasoning and the development of more robust models.

RealHiTBench: A Comprehensive Realistic Hierarchical Table Benchmark for Evaluating LLM-Based Table Analysis

While large language models (LLMs) have shown considerable promise in code generation, real-world software development demands advanced repository-level reasoning. This includes understanding dependencies, project structures, and managing multi-file changes. However, the ability of LLMs to effectively comprehend and handle complex code repositories has yet to be fully explored. To address these challenges, we introduce a hierarchical benchmark designed to evaluate repository dependency understanding(DependEval) for LLMs. The benchmark is based on 2683 repositories collected from real-world websites. It evaluates models on three core tasks: Dependency Recognition, Repository Construction, and Multi-file Editing, across 8 programming languages from actual code repositories. Our evaluation of over 25 LLMs reveals substantial performance gaps and provides valuable insights into repository-level code understanding.

DependEval: Benchmarking LLMs for Repository Dependency Understanding

ICD Coding aims to assign a wide range of medical codes to a medical text document, which is a popular and challenging task in the healthcare domain. To alleviate the problems of long-tail distribution and the lack of annotations of code-specific evidence, many previous works have proposed incorporating code knowledge to improve coding performance. However, existing methods often focus on a single type of knowledge and design specialized modules that are complex and incompatible with each other, thereby limiting their scalability and effectiveness. To address this issue, we propose GKI-ICD, a novel, general knowledge injection framework that integrates three key types of knowledge, namely ICD Description, ICD Synonym, and ICD Hierarchy, without specialized design of additional modules. The comprehensive utilization of the above knowledge, which exhibits both differences and complementarity, can effectively enhance the ICD coding performance. Extensive experiments on existing popular ICD coding benchmarks demonstrate the effectiveness of GKI-ICD, which achieves the state-of-the-art performance on most evaluation metrics. Code is available at https://github.com/xuzhang0112/GKI-ICD.

A General Knowledge Injection Framework for ICD Coding

While large language models (LLMs) show great potential in temporal reasoning, most existing work focuses heavily on enhancing performance, often neglecting the explainable reasoning processes underlying the results. To address this gap, we introduce a comprehensive benchmark covering a wide range of temporal granularities, designed to systematically evaluate LLMs' capabilities in explainable temporal reasoning. Furthermore, our findings reveal that LLMs struggle to deliver convincing explanations when relying solely on textual information. To address challenge, we propose GETER, a novel structure-aware generative framework that integrates Graph structures with text for Explainable TEmporal Reasoning. Specifically, we first leverage temporal knowledge graphs to develop a temporal encoder that captures structural information for the query. Subsequently, we introduce a structure-text prefix adapter to map graph structure features into the text embedding space. Finally, LLMs generate explanation text by seamlessly integrating the soft graph token with instruction-tuning prompt tokens. Experimental results indicate that GETER achieves state-of-the-art performance while also demonstrating robust generalization capabilities. Our dataset and code are available at https://github.com/carryTatum/GETER.

Downloads

Next from ACL 2025

Visibility as Survival: Generalizing NLP for Native Alaskan Language Identification

Similar lecture

Reasoning for Translation: Comparative Analysis of Chain-of-Thought and Tree-of-Thought Prompting for LLM Translation