Austria

In this paper, we identify a critical problem, &quot;lost-in-retrieval&quot;, in retrieval-augmented multi-hop question answering (QA): the key entities are missed in LLMs&#39; sub-question decomposition. &quot;Lost-in-retrieval&quot; significantly degrades the retrieval performance, which disrupts the reasoning chain and leads to the incorrect answers. To resolve this problem, we propose a progressive retrieval and rewriting method, namely ChainRAG, which sequentially handles each sub-question by completing missing key entities and retrieving relevant sentences from a sentence graph for answer generation. Each step in our retrieval and rewriting process builds upon the previous one, creating a seamless chain that leads to accurate retrieval and answers. Finally, all retrieved sentences and sub-question answers are integrated to generate a comprehensive answer to the original question. We evaluate ChainRAG on three multi-hop QA datasets—MuSiQue, 2Wiki, and HotpotQA—using three large language models: GPT4o-mini, Qwen2.5-72B, and GLM-4-Plus. Empirical results demonstrate that ChainRAG consistently outperforms baselines in both effectiveness and efficiency.

ACL 2025

Mitigating Lost-in-Retrieval Problems in Retrieval Augmented Multi-Hop Question Answering

few-shot generation

text-to-text generation

retrieval-augmented generation

### 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)

In this paper, we identify a critical problem, "lost-in-retrieval", in retrieval-augmented multi-hop question answering (QA): the key entities are missed in LLMs' sub-question decomposition. "Lost-in-retrieval" significantly degrades the retrieval performance, which disrupts the reasoning chain and leads to the incorrect answers. To resolve this problem, we propose a progressive retrieval and rewriting method, namely ChainRAG, which sequentially handles each sub-question by completing missing key entities and retrieving relevant sentences from a sentence graph for answer generation. Each step in our retrieval and rewriting process builds upon the previous one, creating a seamless chain that leads to accurate retrieval and answers. Finally, all retrieved sentences and sub-question answers are integrated to generate a comprehensive answer to the original question. We evaluate ChainRAG on three multi-hop QA datasets—MuSiQue, 2Wiki, and HotpotQA—using three large language models: GPT4o-mini, Qwen2.5-72B, and GLM-4-Plus. Empirical results demonstrate that ChainRAG consistently outperforms baselines in both effectiveness and efficiency.

poster

The quality of Supervised Fine-Tuning (SFT) data plays a critical role in enhancing the conversational capabilities of Large Language Models (LLMs). However, the availability of high-quality human-annotated SFT data has become a significant bottleneck for LLMs, necessitating a greater reliance on synthetic training data. In this work, we introduce Condor, a two-stage synthetic data generation framework that incorporates World Knowledge Trees and Self-Reflection Refinement to produce high-quality SFT data at scale. Our experimental results demonstrate that a base model fine-tuned on only 20K Condor-generated samples achieves superior performance compared to instruct model trained with RLHF. The additional refinement stage in Condor further enables iterative self-improvement for LLMs at various scales (up to 72B), validating the effectiveness of our approach. Furthermore, our investigation into the scaling of synthetic data in post-training reveals substantial unexplored potential for performance improvements, opening promising avenues for future research.

Condor: Enhance LLM Alignment with Knowledge-Driven Data Synthesis and Refinement

Multi-agent collaborative tasks exhibit exceptional capabilities in natural language applications and generation. By prompting agents to assign clear roles, it is possible to facilitate cooperation and achieve complementary capabilities among LLMs. A common strategy involves adopting a relatively general role assignment mechanism, such as introducing a ``judge'' or a ``summarizer''. However, these approaches lack task-specific role customization based on task characteristics. Another strategy involves decomposing the task based on domain knowledge and task characteristics, followed by assigning appropriate roles according to LLMs' respective strengths, such as programmers and testers. However, in some given tasks, obtaining domain knowledge related to task characteristics and getting the strengths of different LLMs is hard. To solve these problems, we propose a Multi-LLM Cooperation (MLC) framework with automatic role assignment capabilities. The core idea of the MLC is to initialize role assignments randomly and then allow the role embeddings to be learned jointly with the downstream task. To capture the state transitions of multiple LLMs during turn-based speaking, the role embedding is sequence-aware. At the same time, to avoid role convergence, the role differentiation module in MLC encourages behavioral differentiation between LLMs while ensuring the LLM team consistency, guiding different LLMs to develop complementary strengths from the optimization level. Our experiments on seven datasets demonstrate that MLC significantly enhances collaboration and expertise, which collaboratively addresses multi-agent tasks.

Advancing Collaborative Debates with Role Differentiation through Multi-Agent Reinforcement Learning

Critical text assessment is at the core of many expert activities, such as fact-checking, peer review, and essay grading. Yet, existing work treats critical text assessment as a black box problem, limiting interpretability and human-AI collaboration. To close this gap, we introduce Structured Reasoning in Critical Text Assessment (STRICTA), a novel specification framework to model text assessment as an explicit, step-wise reasoning process. STRICTA breaks down the assessment into a graph of interconnected reasoning steps drawing on causality theory (Pearl, 1995). This graph is populated based on expert interaction data and used to study the assessment process and facilitate human-AI collaboration. We formally define STRICTA and apply it in a study on biomedical paper assessment, resulting in a dataset of over 4000 reasoning steps from roughly 40 biomedical experts on more than 20 papers. We use this dataset to empirically study expert reasoning in critical text assessment, and investigate if LLMs are able to imitate and support experts within these workflows. The resulting tools and datasets pave the way for studying collaborative expert-AI reasoning in text assessment, in peer review and beyond.

STRICTA: Structured Reasoning in Critical Text Assessment for Peer Review and Beyond

To meet the demands of content moderation, online platforms have resorted to automated systems. Newer forms of real-time engagement ($\textit{e.g.}$, users commenting on live streams) on platforms like Twitch exert additional pressures on the latency expected of such moderation systems. Despite their prevalence, relatively little is known about the effectiveness of these systems. In this paper, we conduct an audit of Twitch's automated moderation tool ($\texttt{AutoMod}$) to investigate its effectiveness in flagging hateful content. For our audit, we create streaming accounts to act as siloed test beds, and interface with the live chat using Twitch's APIs to send over $107,000$ comments collated from $4$ datasets. We measure $\texttt{AutoMod}$'s accuracy in flagging blatantly hateful content containing misogyny, racism, ableism and homophobia. Our experiments reveal that a large fraction of hateful messages, up to $94$% on some datasets, $\text{\textit{bypass moderation}}$. Contextual addition of slurs to these messages results in $100$% removal, revealing $\texttt{AutoMod}$'s reliance on slurs as a hate signal. We also find that contrary to Twitch's community guidelines, $\texttt{AutoMod}$ blocks up to $89.5$% of benign examples that use sensitive words in pedagogical or empowering contexts. Overall, our audit points to large gaps in $\texttt{AutoMod}$'s capabilities and underscores the importance for such systems to understand context effectively.

Silencing Empowerment, Allowing Bigotry: Auditing the Moderation of Hate Speech on Twitch

We propose EdiText, a controllable text editing method that modifies the reference text to desired attributes at various scales. We integrate an SDEdit-based editing technique that allows for broad adjustments in the degree of text editing. Additionally, we introduce a novel fine-level editing method based on self-conditioning, which allows subtle control of reference text. While being capable of editing on its own, this fine-grained method, integrated with the SDEdit approach, enables EdiText to make precise adjustments within the desired range. EdiText demonstrates its controllability to robustly adjust reference text at a broad range of levels across various tasks, including toxicity control and sentiment control.

EdiText: Controllable Coarse-to-Fine Text Editing with Diffusion Language Models

Recent advancements have demonstrated the advantage of converting pretrained large language models into powerful text encoders by enabling bidirectional attention in transformer layers. However, existing methods often require extensive training on large-scale datasets, posing challenges in low-resource, domain-specific scenarios. In this work, we show that a pretrained large language model can be converted into a strong text encoder without additional training. We first conduct a comprehensive empirical study to investigate different conversion strategies and identify the impact of the attention sink phenomenon on the performance of converted encoder models. Based on our findings, we propose a novel approach that enables bidirectional attention and suppresses the attention sink phenomenon, resulting in superior performance. Extensive experiments on multiple domains demonstrate the effectiveness of our approach. Our work provides new insights into the training-free conversion of text encoders in low-resource scenarios and contributes to the advancement of domain-specific text representation generation. Our code is available at https://github.com/bigai-nlco/Look-Both-Ways-and-No-Sink.

Look Both Ways and No Sink: Converting LLMs into Text Encoders without Training

The NLP community has converged on considering disagreement in annotation (or human label variation, HLV) as a constitutive feature of subjective tasks. This paper makes a further step by investigating the relationship between HLV and model uncertainty, and the impact of linguistic features of the items on both. We focus on the identification of moral foundations (e.g., care, fairness, loyalty) and human values (e.g., be polite, be honest) in text. We select three standard datasets and proceed into two steps. First, we focus on HLV and analyze the linguistic features (complexity, polarity, pragmatic phenomena, lexical choices) that correlate with HLV. Next, we proceed to uncertainty and its relationship to HLV. We experiment with RoBERTa and Flan-T5 in a number of training setups and evaluation metrics that test the calibration of uncertainty to HLV and its relationship to performance beyond majority vote; next, we analyze the impact of linguistic features on uncertainty. We find that RoBERTa with soft loss is better calibrated to HLV, and we find alignment between calibrated models and humans in the features (textual complexity and polarity) triggering variation.

Mining the uncertainty patterns of humans and models in the annotation of moral foundations and human values

Embeddings-as-a-Service (EaaS) is a service offered by large language model (LLM) developers to supply embeddings generated by LLMs. Previous research suggests that EaaS is prone to imitation attacks—attacks that clone the underlying EaaS model by training another model on the queried embeddings. As a result, EaaS watermarks are introduced to protect the intellectual property of EaaS providers. In this paper, we first show that existing EaaS watermarks can be removed by paraphrasing when attackers clone the model. Subsequently, we propose a novel watermarking technique that involves linearly transforming the embeddings, and show that it is empirically and theoretically robust against paraphrasing.

WET: Overcoming Paraphrasing Vulnerabilities in Embeddings-as-a-Service with Linear Transformation Watermarks

According to psychological and neuroscientific research, a high-stress environment can restrict attentional resources and intensify negative affect, thereby impairing the ability to understand emotions. Furthermore, demographic attributes such as race, gender, and age group have been repeatedly reported to cause significant differences in emotional expression and recognition. This study is the first to systematically verify whether these psychological findings observed in humans also apply to the latest Large Vision Language Models (LVLMs). We constructed low-stress versus high-stress environments and generated an image dataset (a total of 540 images) that combines race, gender, and age group. Based on this, we applied the Pretend prompt technique to induce LVLMs to interpret others’ emotions from the standpoint of the assigned environment and persona. An analysis of the models’ emotional understanding ability, using EQ-Bench-based metrics, revealed that (1) under high-stress environments, the accuracy of emotion understanding significantly declined in most LVLMs, and (2) performance disparities were confirmed across race, gender, and age group. These findings suggest that the effects of high-stress and demographic attributes identified in human research may also be reflected in LVLMs.

Does the Emotional Understanding of LVLMs Vary Under High-Stress Environments and Across Different Demographic Attributes?

Fine-tuning large language models (LLMs) on task-specific data is essential for their effective deployment. As dataset sizes grow, efficiently selecting optimal subsets for training becomes crucial to balancing performance and computational costs. Traditional data selection methods often require fine-tuning a scoring model on the target dataset, which is time-consuming and resource-intensive, or rely on heuristics that fail to fully leverage the model's predictive capabilities. To address these challenges, we propose Data Whisperer, an efficient, training-free, attention-based method that leverages few-shot in-context learning with the model to be fine-tuned. Comprehensive evaluations were conducted on both raw and synthetic datasets across diverse tasks and models. Notably, Data Whisperer achieves superior performance compared to the full GSM8K dataset on the Llama-3-8B-Instruct model, using just 10% of the data, and outperforms existing methods with a 3.1-point improvement and a 7.4× speedup.

Downloads

Next from ACL 2025

Condor: Enhance LLM Alignment with Knowledge-Driven Data Synthesis and Refinement

Similar lecture

GeAR: Graph-enhanced Agent for Retrieval-augmented Generation