REASONING CAPABILITIES

WorkForceAgent-R1: Incentivizing Reasoning Capability in LLM-based Web Agents via Reinforcement Learning (2026.findings-eacl)

Challenge:	Recent advances in large language models have improved the detection of non-compliant content, but critical gaps persist in fine-grained understanding, explainability, and generalization.
Approach:	They propose a framework that combines active reinforcement learning, fine-grained violation understanding and progressive multi-stage training.
Outcome:	The proposed framework outperforms general-purpose LLMs and specialized models in fine-grained violation understanding, explainability, and generalization.

Copied to clipboard

Yuchen Zhuang, Di Jin, Jiaao Chen, Wenqi Shi, Hanrui Wang, Chao Zhang

Distilling Text Style Transfer With Self-Explanation From LLMs (2024.naacl-srw)

Challenge:	Existing web agents relying on supervised fine-tuning struggle with generalization and robustness due to insufficient reasoning capabilities when handling the inherently dynamic nature of web interactions.
Approach:	They propose a large language model-empowered web agent that trains using a rule-based reinforcement learning framework to enhance single-step reasoning and planning for business-oriented web navigation tasks.
Outcome:	The proposed agent outperforms baseline LLM-based agents on the WorkArena benchmark by 10.26–16.59%.

Copied to clipboard

Chiyu Zhang, Honglong Cai, Yuezhang Li, Yuexin Wu, Le Hou, Muhammad Abdul-Mageed

RoT: Enhancing Table Reasoning with Iterative Row-Wise Traversals (2025.emnlp-main)

Challenge:	Text Style Transfer (TST) aims to alter the style of text while preserving its core content.
Approach:	They propose a framework that leverages large language models alongside chain-of-thought prompting to facilitate TST.
Outcome:	The proposed framework surpasses supervised fine-tuning and knowledge distillation methods in low-resource settings.

Copied to clipboard

Xuanliang Zhang, Dingzirui Wang, Keyan Xu, Qingfu Zhu, Wanxiang Che

SportReason: Evaluating Retrieval-Augmented Reasoning across Tables and Text for Sports Question Answering (2025.emnlp-main)

Challenge:	Recent advances in reasoning large language models (RLLMs) have significantly enhanced reasoning capabilities, leading to brilliant performance on table reasoning.
Approach:	They propose a method which performs iterative row-wise table traversal, allowing for reasoning extension and reflection-based refinement at each traversal.
Outcome:	Experiments show that the proposed method outperforms RLLMs on WikiTableQuestions and TableBench by 4.3% and achieves state-of-the-art results with comparable models.

Copied to clipboard

Kaiyue Feng, Siyue Zhang, Bingsen Chen, Yilun Zhao, Chen Zhao

Personal Large Language Model Agents: A Case Study on Tailored Travel Planning (2024.emnlp-industry)

Challenge:	Existing benchmarks for retrieval-augmented reasoning on numerical sports questions focus on one or two evidence units.
Approach:	They propose a benchmark for retrieval-augmented reasoning on numerical sports questions . they evaluate existing retrievers and rerankers, along with agentic Retrieval-Augmented Generation systems.
Outcome:	The proposed benchmarks focus on the sports domain because it offers rich multi-modal resources.

Copied to clipboard

Harmanpreet Singh, Nikhil Verma, Yixiao Wang, Manasa Bharadwaj, Homa Fashandi, Kevin Ferreira, Chul Lee

ArgGen: Prompting Text Generation Models for Document-Level Event-Argument Aggregation (2022.findings-aacl)

Challenge:	Large Language Models (LLMs) are becoming more autonomous and capable of handling real-world tasks through their access to tools, various planning strategies, and memory, referred to as LLM agents.
Approach:	They introduce a personalized version of TravelPlanner and establish baselines for personal LLM agents by comparing generic and personal plans.
Outcome:	The proposed model encapsulates user-related information, preferences, and personal concepts and provides baselines for personal LLM agents.

Copied to clipboard

Debanjana Kar, Sudeshna Sarkar, Pawan Goyal

Preference-Aware Memory Update for Long-Term LLM Agents (2026.findings-acl)

Challenge:	Existing discourse-level information extraction tasks are extractive in nature, but extracting information from larger bodies of discourse-like documents requires more natural language understanding and reasoning capabilities.
Approach:	They propose a conditional text generation approach which generates consolidated event-arguments at a document-level with minimal loss of information.
Outcome:	The proposed approach generates document-level argument spans in a low-resource and zero-shot setting and can be leveraged in other related multilingual text generation tasks.

Copied to clipboard

Haoran Sun, Zekun Zhang, Shaoning Zeng

Dialogue Systems & Conversational Agents

Challenge:	Existing methods for integrating long-term memory do not provide dynamic and personalized memory refinement.
Approach:	They propose a long-term memory update mechanism that enables dynamic and personalized memory refinement.
Outcome:	The proposed mechanism improves the performance of LLM-based agents in five tasks.

Mitigating Hallucination by Integrating Knowledge Graphs into LLM Inference – a Systematic Literature Review (2025.acl-srw)

Copied to clipboard

Robin Wagner, Emanuel Kitzelmann, Ingo Boersch

Structured Data in NLP

Challenge:	Large Language Models (LLMs) have made significant progress on different language tasks, but they tend to "hallucinate" plausible but factually incorrect answers.
Approach:	They propose to integrate knowledge graphs (KGs) into LLM inference to reduce hallucinations by searching online and applying a selection process.
Outcome:	The proposed integration improves performance on benchmark datasets and also to mitigate hallucinations.

ThinkSum: Probabilistic reasoning over sets using large language models (2023.acl-long)

Copied to clipboard

Batu Ozturkler, Nikolay Malkin, Zhen Wang, Nebojsa Jojic

Teaching-Inspired Integrated Prompting Framework: A Novel Approach for Enhancing Reasoning in Large Language Models (2025.coling-industry)

Challenge:	Large language models (LLMs) have a substantial capacity for high-level analogical reasoning, but they fail in scenarios that require reasoning over multiple objects or facts and making sequences of logical deductions.
Approach:	They propose a two-stage probabilistic inference paradigm, ThinkSum, which reasons over sets of objects or facts in a structured manner.
Outcome:	The proposed paradigm improves on the BIG-bench suite of evaluation tasks.

Copied to clipboard

Wenting Tan, Dongxiao Chen, Jieting Xue, Zihao Wang, Taijie Chen

MMC: Advancing Multimodal Chart Understanding with Large-scale Instruction Tuning (2024.naacl-long)

Challenge:	Large Language Models (LLMs) exhibit impressive performance across various domains but struggle with arithmetic reasoning tasks.
Approach:	They propose a Teaching-Inspired Integrated Prompting Framework which emulates the instructional process of a teacher guiding students.
Outcome:	The proposed framework improves reasoning accuracy on nine benchmarks.

Copied to clipboard

Fuxiao Liu, Xiaoyang Wang, Wenlin Yao, Jianshu Chen, Kaiqiang Song, Sangwoo Cho, Yaser Yacoob, Dong Yu

Is ChatGPT a General-Purpose Natural Language Processing Task Solver? (2023.emnlp-main)

Challenge:	Existing large language models have limited ability to perform tasks effectively.
Approach:	They propose a large-scale multimodal chart instruction dataset with 600k instances supporting diverse tasks and chart types.
Outcome:	The proposed LMM achieves state-of-the-art performance on existing chart QA benchmarks.

Copied to clipboard

Chengwei Qin, Aston Zhang, Zhuosheng Zhang, Jiaao Chen, Michihiro Yasunaga, Diyi Yang

RV-Syn: Rational and Verifiable Mathematical Reasoning Data Synthesis based on Structured Function Library (2026.findings-eacl)

Challenge:	Recent advances in scale have enabled large language models to perform NLP tasks zero-shot . however, it is not known whether ChatGPT can serve as a generalist model that can perform many NLP jobs zero- shot.
Approach:	They empirically evaluate ChatGPT's zero-shot learning ability on 20 popular NLP datasets . they find it performs well on many tasks favoring reasoning abilities .
Outcome:	The proposed model can perform many NLP tasks zero-shot without adaptation on downstream data.

Copied to clipboard

Jiapeng Wang, Jinhao Jiang, Zhiqiang Zhang, Jun Zhou, Xin Zhao

Making Long-Context Language Models Better Multi-Hop Reasoners (2024.acl-long)

Challenge:	Existing methods for generating high-quality reasoning data are limited in quality and availability.
Approach:	They propose a method that constructs mathematical operations and generates verifiable graphs that are back-translated into complex problems.
Outcome:	The proposed method achieves a 6.3% performance gain over existing methods on LLaMA-3-8B and outperforms others with only half the training data (50k vs. 100k).

Copied to clipboard

Yanyang Li, Shuo Liang, Michael Lyu, Liwei Wang

Learning from Diverse Reasoning Paths with Routing and Collaboration (2025.emnlp-main)

Challenge:	Recent advances in long-context modeling have enhanced language models for complex tasks, but they struggle with multi-hop reasoning and noisy contexts.
Approach:	They propose an approach that prompts LMs to supply attributions for each assertion during reasoning.
Outcome:	The proposed model achieves competitive performance on multi-hop reasoning benchmarks, closely paralleling proprietary LMs such as ChatGPT and Claude-instant.

Copied to clipboard

Zhenyu Lei, Zhen Tan, Song Wang, Yaochen Zhu, Zihan Chen, Yushun Dong, Jundong Li

PaD: Program-aided Distillation Can Teach Small Models Reasoning Better than Chain-of-thought Fine-tuning (2024.naacl-long)

Challenge:	Recent studies suggest that the reasoning abilities of large language models (LLMs) grows with model size and pre-training data.
Approach:	They propose to combine quality filtering, conditional routing, and cooperative peer teaching to transfer knowledge from powerful teacher models to compact and transparent students.
Outcome:	Experiments show that QR-Distill is superior to traditional methods.

Copied to clipboard

Xuekai Zhu, Biqing Qi, Kaiyan Zhang, Xinwei Long, Zhouhan Lin, Bowen Zhou

An Electoral Approach to Diversify LLM-based Multi-Agent Collective Decision-Making (2024.emnlp-main)

Challenge:	Large language models excel in various tasks, but their huge size and inaccessibility of parameters present challenges for practical deployment.
Approach:	They propose to use CoT data to distill task-specific ability from large language models to smaller models . they use reasoning programs to suppress errors in distilled data and improve distillation quality .
Outcome:	The proposed model outperforms LLMs on arithmetic reasoning, symbolic reasoning, and general ability.

Copied to clipboard

Xiutian Zhao, Ke Wang, Wei Peng

Can LLMs Learn from Previous Mistakes? Investigating LLMs’ Errors to Boost for Reasoning (2024.acl-long)

Challenge:	Recent advances in large language models have sparked interest in collaborative LLM agents.
Approach:	They propose to integrate various ordinal preferential voting mechanisms into LLMs to improve reasoning capabilities and robustness.
Outcome:	The proposed method improves reasoning capabilities and robustness of leading LLMs without complex system designs.

Copied to clipboard

Yongqi Tong, Dawei Li, Sizhe Wang, Yujia Wang, Fei Teng, Jingbo Shang

Low-Resource NLP

Challenge:	Recent studies have shown the benefits to LLMs from fine-tuning golden-standard Chain-of-Thought rationales or using them as correct examples in few-shot prompting.
Approach:	They propose a new benchmark to test the effectiveness of large language models by leveraging errors to enhance reasoning capabilities.
Outcome:	The proposed methods can be used to fine-tune models in correct and incorrect domains, rather than tuning models to learn ground truth in traditional methods.

VoCoT: Unleashing Visually Grounded Multi-Step Reasoning in Large Multi-Modal Models (2025.naacl-long)

Copied to clipboard

Zejun Li, Ruipu Luo, Jiwen Zhang, Minghui Qiu, Xuanjing Huang, Zhongyu Wei

Reasoning with Trees: Faithful Question Answering over Knowledge Graph (2025.coling-main)

Challenge:	Despite the impressive capabilities of large multi-modal models, their effectiveness in handling complex tasks has been limited by the prevailing singlestep reasoning paradigm.
Approach:	They propose a visuallygrounded object-centric Chain-of-Thought reasoning framework for LMMs that is based on a multi-modal interleaved and aligned representation of object concepts.
Outcome:	The proposed model outperforms SOTA models in CLEVR and EmbSpatial benchmarks.

Copied to clipboard

Tiesunlong Shen, Jin Wang, Xuejie Zhang, Erik Cambria

Knowledge Representation

Challenge:	Recent advances in large language models (LLMs) have shown remarkable progress in reasoning capabilities, yet they still face challenges in complex, multi-step reasoning tasks.
Approach:	They propose a framework that synergistically integrates LLMs with knowledge graphs (KGs) to enhance reasoning performance and interpretability.
Outcome:	The proposed framework outperforms existing state-of-the-art methods on two benchmark KGQA datasets and improves on the MCTS process.

Entropy-Gated Branching for Efficient Test-Time Reasoning (2026.eacl-long)

Copied to clipboard

Xianzhi Li, Ethan Callanan, Abdellah Ghassel, Xiaodan Zhu

Prejudge-Before-Think: Enhancing Large Language Models at Test-Time by Process Prejudge Reasoning (2025.findings-emnlp)

Challenge:	Empirical results show that branching at low uncertainty points can improve reasoning capabilities of large language models . however, these methods require substantially more computational resources, causing errors in high-stakes domains .
Approach:	They propose an inference technique that selectively expands prediction sequences at points of high uncertainty.
Outcome:	Empirical results show that the proposed method improves accuracy by 22.6% over standard inference while operating 31%-75% faster across math benchmarks.

Copied to clipboard

Jianing Wang, Jin Jiang, Yang Liu, Mengdi Zhang, Xunliang Cai

RA-ISF: Learning to Answer and Understand from Retrieval Augmentation via Iterative Self-Feedback (2024.findings-acl)

Challenge:	Experimental results from competition-level complex reasoning demonstrate that bootstrapping with process prejudge can significantly enhance the reasoning ability of LLMs.
Approach:	They propose a new process prejudge strategy for LLM reasoning that bootstraps with process prejudgment .
Outcome:	The proposed method can be bootstrapped with process prejudge in LLM reasoning . it allows the model to anticipate errors rather than relying on trial and error.

Copied to clipboard

Yanming Liu, Xinyue Peng, Xuhong Zhang, Weihao Liu, Jianwei Yin, Jiannan Cao, Tianyu Du

Information Retrieval

Challenge:	Large language models (LLMs) have demonstrated excellent performance in numerous tasks but the parameterized knowledge stored within LLMs may be incomplete and hard to incorporate up-to-date knowledge.
Approach:	They propose a framework that iteratively decomposes tasks and processes them in three submodules to enhance the model’s problem-solving capabilities.
Outcome:	The proposed method outperforms existing benchmarks on GPT3.5, Llama2 and other large language models significantly enhancing factual reasoning capabilities and reducing hallucinations.

Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs (2026.findings-acl)

Copied to clipboard

Hongyuan Yuan, Xinran He, Run Shao, Bolei He, Xianwei Xue, Mengke Chen, Qiutong Pan, Haiwei Wang, Haifeng Li

Structured Data in NLP

Challenge:	Extending CoT through RL can induce undesirable thinking patterns such as overthinking . prior work has focused on inefficient reflection, which manifests in two problematic patterns: Indiscriminate Reflection and Repetitive Reflectione .
Approach:	They propose a graph-based approach to optimize CoT by pruning each linear CoT into a directed acyclic graph with explicit dependency edges.
Outcome:	The proposed approach reduces the average reasoning tokens by 42% while maintaining or improving accuracy.

Disentangling Biased Knowledge from Reasoning in Large Language Models via Machine Unlearning (2025.acl-long)

Copied to clipboard

Zheyuan Liu, Suraj Maharjan, Fanyou Wu, Rahil Parikh, Belhassen Bayar, Srinivasan H. Sengamedu, Meng Jiang

TinyThinker: Distilling Reasoning through Coarse-to-Fine Knowledge Internalization with Self-Reflection (2025.naacl-long)

Challenge:	Existing approaches to disentangle biased knowledge from reasoning are sub-optimal . entangled data makes curation difficult, leading to inclusion of sensitive, toxic data.
Approach:	They propose a framework that selectively removes biased knowledge while preserving reasoning abilities.
Outcome:	The proposed framework improves fairness accuracy by 14.7% and reasoning performance by 62.6% across multiple LLMs.

Copied to clipboard

Shengmin Piao, Sanghyun Park

Thinking about how to extract: Energizing LLMs’ emergence capabilities for document-level event argument extraction (2024.findings-acl)

Challenge:	Large Language Models exhibit impressive reasoning capabilities across diverse tasks . direct training on synthesized reasoning data may lead to superficial imitation of reasoning process, authors argue .
Approach:	They propose a framework that introduces a three-stage process that incrementally guides the student model through the reasoning process, progressively refining knowledge from coarse to fine granularity.
Outcome:	The proposed framework achieves superior performance on commonsense reasoning benchmarks and can be extended to other knowledge-intensive reasoning tasks.

Copied to clipboard

Kai Shuang, Zhouji Zhouji, Wang Qiwei, Jinyu Guo

Information Extraction & Text Mining

Challenge:	Existing models for document-level event argument extraction (D-EAE) lack key feature forgetting and cross-event argument confusion.
Approach:	They propose a document-level event argument extraction method based on guided summarization and reasoning that leverages the emergence capabilities of large language models to highlight key event information.
Outcome:	The proposed method outperforms baseline models by 1.3% F1 and 1.6% F1 on WIKIEVENTS and RAMS.

Teaching Small Language Models Reasoning through Counterfactual Distillation (2024.emnlp-main)

Copied to clipboard

Tao Feng, Yicheng Li, Li Chenglin, Hao Chen, Fei Yu, Yin Zhang

The Self-Improvement Paradox: Can Language Models Bootstrap Reasoning Capabilities without External Scaffolding? (2025.findings-acl)

Challenge:	Large language models (LLMs) have demonstrated remarkable performance in a wide range of downstream tasks.
Approach:	They propose a counterfactual distillation framework that leverages LLMs to generate high-quality counterfacts and utilizes multi-view CoT to enhance the diversity of reasoning samples.
Outcome:	The proposed framework enhances reasoning capabilities of large language models and is more robust to OOD data.

Copied to clipboard

Yutao Sun, Mingshuai Chen, Tiancheng Zhao, Ruochen Xu, Zilun Zhang, Jianwei Yin

Towards Debiasing Fact Verification Models (D19-1)

Challenge:	Existing approaches to self-improvement rely on external supervision signals in the form of seed data and/or assistance from third-party models.
Approach:	They propose a framework for generating high-quality synthetic question-answer data in a fully autonomous manner.
Outcome:	The proposed framework generates high-quality synthetic question-answer data in a fully autonomous manner.

Copied to clipboard

Tal Schuster, Darsh Shah, Yun Jie Serene Yeo, Daniel Roberto Filizzola Ortiz, Enrico Santus, Regina Barzilay

Breaking the Reasoning Barrier A Survey on LLM Complex Reasoning through the Lens of Self-Evolution (2025.findings-acl)

Challenge:	Prior research has shown that data collection methods that use crowdsourcing introduce idiosyncratic biases that impact performance in unexpected ways.
Approach:	They propose a method to regularize the training data to avoid idiosyncrasies in the datasets that are used for fact verification.
Outcome:	The proposed model outperforms the existing model on the FEVER dataset, achieving 61.7% of the baseline.

Copied to clipboard

Tao He, Hao Li, Jingchang Chen, Runxuan Liu, Yixin Cao, Lizi Liao, Zihao Zheng, Zheng Chu, Jiafeng Liang, Ming Liu, Bing Qin

Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot Do (2026.acl-long)

Challenge:	OpenAI's O1 and subsequent projects like DeepSeek R1 have significantly advanced research on complex reasoning in LLMs.
Approach:	They analyze existing reasoning studies from the perspective of self-evolution and summarize O1-like works from open-source projects like DeepSeek R1 and Kimi-k1.5.
Outcome:	The proposed models are based on open-source models and pioneer advanced methodologies like Scaling Reinforcement Learning (RL).

Copied to clipboard

Zhuoran Jin, Kejian Zhu, Hongbang Yuan, Yupu Hao, Pengfei Cao, Yubo Chen, Kang Liu, Jun Zhao

Revisiting Model Interpolation for Efficient Reasoning (2026.acl-long)

Challenge:	Existing open-source models often yield only marginal overall improvements, possibly due to an overemphasis on mathematical reasoning at the expense of broader capabilities.
Approach:	They evaluate 12 multimodal tasks using 14 non-reasoning models and 8 reasoning models.
Outcome:	The proposed method is effective in multimodal reasoning tasks, the authors show . they show that it lacks the ability to maintain deep visual introspection throughout the reasoning process.

Copied to clipboard

Taiqiang Wu, Runming Yang, Tao Liu, Jiahao Wang, Ngai Wong

MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning (2026.acl-long)

Challenge:	Existing models that interpolate weights of two specialized models can be abused for efficient reasoning.
Approach:	They propose to merge two specialized models and create a model that combines efficiency and efficiency.
Outcome:	The proposed method outperforms existing models on efficiency and effectiveness.

Copied to clipboard

Yizhe Zeng, Wei Zhang, Yunpeng Li, Juxin Xiao, Xiao Wang, Yuling Liu

UBench: Benchmarking Uncertainty in Large Language Models with Multiple Choice Questions (2025.findings-acl)

Challenge:	Existing CoT backdoor attacks manipulate intermediate reasoning steps to steer the model toward incorrect answers, but these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses.
Approach:	They propose a backdoor attack that exploits the model's post-output space to preserve clean CoTs while selectively steering the final answer toward a specific target.
Outcome:	Experiments show that MirageBD achieves over 90% success rate across four datasets and five models with a poison ratio of only 5%.

Copied to clipboard

Xunzhi Wang, Zhuowei Zhang, Gaonan Chen, Qiongyu Li, Bitong Luo, Zhixin Han, Haotian Wang, Zhiyu Li, Hang Gao, Mengting Hu

SynDARin: Synthesising Datasets for Automated Reasoning in Low-Resource Languages (2025.coling-main)

Challenge:	Existing methods for benchmarking the uncertainty of large language models face challenges . existing methods require internal model access, additional training, or high computational costs .
Approach:	They propose a new benchmark for evaluating the uncertainty of large language models based on confidence intervals . UBench encompasses 11,978 multiple choice questions spanning knowledge, language, understanding, and reasoning capabilities.
Outcome:	The proposed method outperforms existing methods for benchmarking the uncertainty of large language models.

Copied to clipboard

Gayane Ghazaryan, Erik Arakelyan, Isabelle Augenstein, Pasquale Minervini

Machine Translation

Challenge:	Question Answering datasets are scarce for languages other than English due to the cost and difficulties of collection and manual annotation.
Approach:	They propose a method for generating and validating QA datasets for low-resource languages . they use English data as context to generate synthetic multiple-choice (MC) question-answer pairs .
Outcome:	The proposed method maintains quality, reduces likelihood of factual errors, and circumvents costly annotation.

IRUEX: A Study on Large Language Models Problem-Solving Skills in Iran’s University Entrance Exam (2025.coling-main)

Copied to clipboard

Hamed Khademi Khaledi, Heshaam Faili

An Empirical Study of Multilingual Reasoning Distillation for Question Answering (2024.emnlp-main)

Challenge:	Recent advances in Large Language Models (LLMs) have profound implications for education.
Approach:	They present a novel multiple-choice educational resource specifically designed to evaluate the performance of Large Language Models (LLMs) they use a dataset that contains 868 questions and 36,485 additional questions .
Outcome:	The IRUEX dataset contains 868 questions and 36,485 additional questions.

Copied to clipboard

Patomporn Payoungkhamdee, Peerat Limkonchotiwat, Jinheon Baek, Potsawee Manakul, Can Udomcharoenchaikit, Ekapol Chuangsuwanich, Sarana Nutanong

Can Large Language Models Grasp Legal Theories? Enhance Legal Reasoning with Insights from Multi-Agent Collaboration (2024.findings-emnlp)

Challenge:	Existing efforts to distill reasoning capabilities have focused mainly on English, leaving multilingual distillation underexplored.
Approach:	They propose a method that incorporates incorrect rationales as additional guidance to improve multilingual reasoning in large language models.
Outcome:	Empirical results show that d-CoT-nR significantly surpasses the baseline, improving accuracy in unseen languages and correctness in step-by-step reasoning.

Copied to clipboard

Weikang Yuan, Junjie Cao, Zhuoren Jiang, Yangyang Kang, Jun Lin, Kaisong Song, Tianqianjin Lin, Pengwei Yan, Changlong Sun, Xiaozhong Liu

Leveraging Structured Information for Explainable Multi-hop Question Answering and Reasoning (2023.findings-emnlp)

Challenge:	Existing studies have found that when LLMs are given criminal facts and legal rules, then asked whether cases constitute a certain charge, they struggle to understand legal theories and perform basic legal reasoning tasks.
Approach:	They propose a task to assess LLMs' understanding of legal theories and reasoning capabilities by using a novel framework: Multi-Agent framework for improving complex legal reasoning capability.
Outcome:	The proposed framework improves LLMs' understanding of legal theories and reasoning abilities in real-world scenarios.

Copied to clipboard

Ruosen Li, Xinya Du

S2-MAD: Breaking the Token Barrier to Enhance Multi-Agent Debate Efficiency (2025.naacl-long)

Challenge:	Neural models, including large language models (LLMs), achieve superior performance on multi-hop question-answering tasks.
Approach:	They propose to use the chain-of-thought mechanism to generate both the reasoning chain and the answer.
Outcome:	Empirical results show that the proposed framework generates more faithful reasoning chains and significantly improves the QA performance on two benchmark datasets.

Copied to clipboard

Yuting Zeng, Weizhe Huang, Lei Jiang, Tongxuan Liu, XiTai Jin, Chen Tianying Tiana, Jing Li, Xiaohua Xu

INREACT: An Inspire-Then-Reinforce Training Framework For Multimodal GUI Agent (2025.findings-emnlp)

Challenge:	Large language models exhibit limitations when handling complex mathematical reasoning and logical inference tasks.
Approach:	They propose a sparsification strategy to reduce token costs within Multi-agent Debate (MAD) this strategy minimizes ineffective exchanges of information and unproductive discussions among agents .
Outcome:	The proposed approach reduces token costs by up to 94.5% while maintaining performance degradation below 2.0%.

Copied to clipboard

Yuanlei Wang, Liuzhou Zhang, Haohao Luo, Ying Shen

Weak-to-Strong Reasoning (2024.findings-emnlp)

Challenge:	Existing multimodal large language models struggle with precise localization of small elements.
Approach:	They propose a multimodal GUI agent framework that unifies observing, thinking, and acting for precise and interpretable decision-making.
Outcome:	The proposed framework unifies observing, thinking, and acting for precise and interpretable decision-making.

Copied to clipboard

Yuqing Yang, Yan Ma, Pengfei Liu

DSG-MCTS: A Dynamic Strategy-Guided Monte Carlo Tree Search for Diversified Reasoning in Large Language Models (2025.emnlp-main)

Challenge:	Existing approaches to supervise large language models (LLMs) exceed human capabilities, but the effectiveness of this approach is still unexplored.
Approach:	They propose a weak-to-strong reasoning framework that enables strong models to refine training data . they use supervised fine-tuning and preference optimization to optimize weak models .
Outcome:	The proposed framework improves the reasoning capabilities of a language model using three weak models.

Copied to clipboard

Rui Ha, Chaozhuo Li, Rui Pu, Litian Zhang, Xi Zhang, Sen Su

Feature Extraction and Steering for Enhanced Chain-of-Thought Reasoning in Language Models (2025.emnlp-main)

Challenge:	Large language models (LLMs) have shown strong potential in complex reasoning tasks, but their performance often degrades, resulting in hallucinations, errors, and logical inconsistencies.
Approach:	They propose a framework that integrates multiple reasoning strategies to expand the reasoning space and a dynamic strategy selection mechanism that adapts to the task context.
Outcome:	The proposed framework outperforms existing state-of-the-art methods on a set of reasoning benchmarks.

Copied to clipboard

Zihao Li, Xu Wang, Yuzhe Yang, Ziyu Yao, Haoyi Xiong, Mengnan Du

PToco: Prefix-based Token-level Collaboration Enhances Reasoning for Multi-LLMs (2025.coling-main)

Challenge:	Large Language Models (LLMs) can solve reasoning and mathematical problems using the Chain-of-Thought technique, but require costly and long CoT data and fine-tuning.
Approach:	They propose a method that uses Sparse Autoencoders to extract interpretable features from vanilla CoT and use them to steer the LLM's internal states.
Outcome:	The proposed method uses Sparse Autoencoders (SAEs) to extract interpretable features from vanilla CoT and steer the LLM's internal states during generation.

Copied to clipboard

Yuang Bian, Yupian Lin, Jingping Liu, Tong Ruan

On the Impact of Fine-Tuning on Chain-of-Thought Reasoning (2025.naacl-long)

Challenge:	Existing approaches to collaboration between multiple Large Language Models (LLMs) rely on highly capable models with strong self-reflection abilities or are limited to models sharing the same tokenizer.
Approach:	They propose a mechanism that enables collaboration among less capable LLMs independent of tokenizer differences.
Outcome:	The proposed mechanism improves performance over individual models and generalizes well across different quantities and sizes of participating models.

Copied to clipboard

Elita Lobo, Chirag Agarwal, Himabindu Lakkaraju

Safe: Enhancing Mathematical Reasoning in Large Language Models via Retrospective Step-aware Formal Verification (2025.acl-long)

Challenge:	Large language models have emerged as powerful tools for general intelligence, showcasing advanced natural language processing capabilities.
Approach:	They propose to use supervised fine-tuning and Quantized Low-Rank Adapters to improve LLMs' task-specific performance to address privacy and safety risks.
Outcome:	The proposed model improves the accuracy of the chain-of-thought reasonings across four datasets and demonstrates that the faithfulness of CoT reasoning decreases.

Copied to clipboard

Chengwu Liu, Ye Yuan, Yichun Yin, Yan Xu, Xin Xu, Zaoyu Chen, Yasheng Wang, Lifeng Shang, Qun Liu, Ming Zhang

Socratic Style Chain-of-Thoughts Help LLMs to be a Better Reasoner (2025.findings-acl)

Challenge:	Chain-of-Thought prompting is a de facto method to elicit reasoning capabilities from large language models (LLMs).
Approach:	They propose a step-aware formal verification framework Safe to address hallucinations in CoT prompting . they propose 'formal step' as a benchmark for step correctness theorem proving with 30,809 formal statements.
Outcome:	The proposed framework shows significant performance improvement while offering interpretable and verifiable evidence.

Copied to clipboard

Jiangbo Pei, Peiyu Liu, Xin Zhao, Aidong Men, Yang Liu

Find-the-Common: A Benchmark for Explaining Visual Patterns from Images (2024.lrec-main)

Challenge:	Synthetic data generation is a promising approach to enhance reasoning capabilities of large language models.
Approach:	They propose a multi-agent debate framework based on the Socratic questioning strategy . they use socratic questions to deepen the thinking process and broaden it to motivate self-reflection .
Outcome:	The proposed framework outperforms existing methods on math and code generation tasks while maintaining affordable costs.

Copied to clipboard

Yuting Shi, Naoya Inoue, Houjing Wei, Yufeng Zhao, Tao Jin

Dynamic Collaboration of Multi-Language Models based on Minimal Complete Semantic Units (2025.emnlp-main)

Challenge:	Recent advances in Instruction-fine-tuned Vision and Language Models (IVLMs) have prompted some studies to analyze the reasoning capabilities of IVLMs.
Approach:	They introduce a vision and language task for Inductive Visual Reasoning that uses common attributes across visual scenes to find common answers.
Outcome:	The proposed model can archive with 48% accuracy on the FTC, compared with state-of-the-art models.

Copied to clipboard

Chao Hao, Zezheng Wang, Yanhua Huang, Ruiwen Xu, Wenzhe Niu, Xin Liu, Zitong Yu

DeepPrune: Parallel Scaling without Inter-trace Redundancy (2026.findings-acl)

Challenge:	Existing methods to enhance reasoning capabilities of language models are expensive and often lack the ability to perform complex reasoning tasks.
Approach:	They propose a token-level multi-model collaboration strategy to enhance reasoning capabilities in language models by selecting the optimal tokens from the next token distributions.
Outcome:	The proposed method is superior to existing methods and will be released soon.

Copied to clipboard

Shangqing Tu, Yaxuan Li, Yushi Bai, Lei Hou, Juanzi Li

Direct Value Optimization: Improving Chain-of-Thought Reasoning in LLMs with Refined Values (2025.emnlp-main)

Challenge:	Parallel scaling is a powerful paradigm to enhance reasoning capabilities in large language models.
Approach:	They propose a framework that enables efficient parallel scaling through dynamic pruning.
Outcome:	The proposed framework achieves token reductions of 65.73% to 88.50% compared to consensus sampling while maintaining competitive accuracy within 3.4 percentage points.

Copied to clipboard

Hongbo Zhang, Han Cui, Guangsheng Bao, Linyi Yang, Jun Wang, Yue Zhang

BioHopR: A Benchmark for Multi-Hop, Multi-Answer Reasoning in Biomedical Domain (2025.findings-acl)

Challenge:	Existing offline preference optimization methods rely on preference labels to optimize large language models.
Approach:	They propose an offline method for enhancing large language models in reasoning tasks that utilizes value signals at individual reasoning steps.
Outcome:	The proposed framework outperforms offline preference optimization techniques by 4% to 6% on math reasoning, commonsense reasoning, and coding tasks.

Copied to clipboard

Yunsoo Kim, Yusuf Abdulle, Honghan Wu

R-PRM: Reasoning-Driven Process Reward Modeling (2025.emnlp-main)

Challenge:	Existing benchmarks for multi-hop reasoning in biomedical domain are lacking . bioHopR provides benchmarks to evaluate multi-step reasoning in structured biomedic knowledge graphs .
Approach:	They propose a benchmark to evaluate multi-hop, multi-answer reasoning in biomedical knowledge graphs.
Outcome:	BioHopR evaluates multi-hop reasoning in biomedical knowledge graphs based on the PrimeKG model . it outperforms proprietary models and open-source biomedal models in 1-hop and 2-hop tasks .

Copied to clipboard

Shuaijie She, Junxiao Liu, Yifeng Liu, Jiajun Chen, Xin Huang, Shujian Huang

MAmmoTH-VL: Eliciting Multimodal Reasoning with Instruction Tuning at Scale (2025.acl-long)

Challenge:	Existing Process Reward Models (PRMs) output evaluation scores directly, limiting both learning efficiency and evaluation accuracy.
Approach:	They propose a Reasoning-Driven Process Reward Modeling (R-PRM) which activates inherent reasoning to enhance process-level evaluation.
Outcome:	The proposed model outperforms baseline models on ProcessBench and PRMBench by 13.9 and 8.5 F1 scores.

Copied to clipboard

Jiawei Guo, Tianyu Zheng, Yizhi Li, Yuelin Bai, Bo Li, Yubo Wang, King Zhu, Graham Neubig, Wenhu Chen, Xiang Yue

Multi-Agent Autonomous Driving Systems with Large Language Models: A Survey of Recent Advances, Resources, and Future Directions (2025.findings-emnlp)

Challenge:	Current instruction-tuning datasets focus on simplistic visual question answering tasks, and provide phrase-level answers without any intermediate rationales.
Approach:	They propose to use open-source multimodal large language models to train MLLMs on a dataset with 12M instruction-response pairs to elicit CoT reasoning.
Outcome:	The proposed model achieves state-of-the-art performance on benchmarks such as MathVerse, MMMU-Pro, and MuirBench, and gains improvements of up to 4% on non-reasoning-based benchmarks.

Copied to clipboard

Yaozu Wu, Dongyuan Li, Yankai Chen, Renhe Jiang, Henry Peng Zou, Wei-Chieh Huang, Yangning Li, Liancheng Fang, Zhen Wang, Philip S. Yu

Logic-Thinker: Teaching Large Language Models to Think more Logically. (2025.findings-emnlp)

Challenge:	Large Language Models (LLMs) are used to assist with driving decisions, but they face limitations in perception and computational demands.
Approach:	They propose a survey of LLM-based multi-agent ADSs and their applications . they analyze agent-human interactions in scenarios where LLM agents engage with humans .
Outcome:	The proposed approach reduces human intervention and improves safety and efficiency.

Copied to clipboard

Chengyao Wen, Qiang Cheng, Shaofei Wang, Zhizhen Liu, Deng Zhao, Lei Liang

ChartM3: A Multi-Stage Code-Driven Pipeline for Constructing Multi-Dimensional and Multi-Step Visual Reasoning Data in Chart Comprehension (2025.findings-emnlp)

Challenge:	Recent Large Reasoning Models (LRMs) have demonstrated the ability to generate long chains of thought (LongCoT) LongCoT still faces challenges such as redundancy and logical incoherence.
Approach:	They propose a neural-symbolic reasoning framework that generates chains of thought . they propose Logic-Thinker, which transforms symbolic solvers into chains of thoughts .
Outcome:	The proposed framework outperforms models fine-tuned with ThinkerCoT on logic reasoning tasks.

Copied to clipboard

Duo Xu, Hao Cheng, Xin Lin, Zhen Xie, Hao Henry Wang

Can Large Language Models Understand Argument Schemes? (2025.findings-acl)

Challenge:	Currently, research on complex chart understanding tasks is limited . a pipeline for visual reasoning datasets addresses these limitations .
Approach:	They propose a code-driven pipeline for generating visual reasoning datasets . pipeline integrates retrieval-augmented generation to retrieve professional chart templates .
Outcome:	The proposed pipeline enhances chart diversity and data quality through model-based evaluation.

Copied to clipboard

Elfia Bezou-Vrakatseli, Oana Cocarascu, Sanjay Modgil

Retrieval Enhanced Feedback via In-context Neural Error-book (2025.emnlp-main)

Challenge:	Argument schemes are stereotypical forms of reasoning that occur in everyday arguments.
Approach:	They propose to use large language models (LLMs) to classify argument schemes based on Walton’s taxonomy to employ formal definitions and LLM-generated descriptions to enhance task instructions.
Outcome:	The proposed models perform well on annotated and automatically generated arguments, and provide insights for advancing reasoning capabilities in computational argumentation.

Copied to clipboard

Jongyeop Hyun, Bumsoo Kim

PizzaCommonSense: A Dataset for Commonsense Reasoning about Intermediate Steps in Cooking Recipes (2024.findings-emnlp)

Challenge:	Existing methods for learning from errors lack a structured framework for analyzing and mitigating errors, especially in Multimodal Large Language Models (MLLMs).
Approach:	They propose a teacher-student framework that systematically structures errors to deliver targeted feedback for multimodal reasoning.
Outcome:	The proposed framework improves inference efficiency, token usage, and scalability by building a query-based structure that prioritizes visual information, diagnoses failure points, and guides corrective actions.

Copied to clipboard

Aissatou Diallo, Antonis Bikakis, Luke Dickens, Anthony Hunter, Rob Miller

MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark (2025.acl-long)

Challenge:	Understanding procedural texts is essential for enabling machines to follow instructions and reason about tasks.
Approach:	They propose a corpus of cooking recipes enriched with descriptions of intermediate steps . they propose enabling machines to follow instructions and reason about tasks .
Outcome:	The proposed model achieves only 26% human-evaluated preference for generations . pizzaCommonsense is a benchmark for the reasoning capabilities of large language models .

Copied to clipboard

Xiang Yue, Tianyu Zheng, Yuansheng Ni, Yubo Wang, Kai Zhang, Shengbang Tong, Yuxuan Sun, Botao Yu, Ge Zhang, Huan Sun, Yu Su, Wenhu Chen, Graham Neubig

What Factors Affect LLMs and RLLMs in Financial Question Answering? (2026.findings-acl)

Challenge:	Recent advances in multimodal large language models have led to progress in tackling complex reasoning tasks that combine textual and visual information.
Approach:	They introduce a robust version of the Massive Multi-discipline Multimodal Understanding and Reasoning (MMMU) benchmark.
Outcome:	The proposed model performs lower on MMMU-Pro than on the previous benchmark, ranging from 16.8% to 26.9%.

Copied to clipboard

Peng Wang, Xuesi Hu, Jiageng Wu, Yuntao Zou, Qiancheng Zhang, Dagang Li

ScienceWorld: Is your Agent Smarter than a 5th Grader? (2022.emnlp-main)

Challenge:	Recent studies have focused on large language models and reasoning large language model (RLLMs) however, there are few studies that explore what methods can fully unlock the performance of LLMs and RLLM in the financial domain.
Approach:	They examine the effects of prompting methods, agentic frameworks, and multilingual alignment methods on financial question-answering tasks.
Outcome:	The results show that prompting methods and agent frameworks improve LLMs' performance . the authors suggest that these frameworks can be used to enhance LLM performance if they are implemented in financial domains.

Copied to clipboard

Ruoyao Wang, Peter Jansen, Marc-Alexandre Côté, Prithviraj Ammanabrolu

Natural Language Interfaces

Challenge:	Existing models cannot reason about or explain learned science concepts in novel contexts, despite transformer-based progress in question-answering and scientific text processing .
Approach:	They propose a benchmark to test agents’ scientific reasoning abilities in a new interactive text environment at the level of a standard elementary school science curriculum.
Outcome:	The proposed model outperforms a model trained for 100k steps in a standard elementary school science curriculum.

Self-Training Large Language Models with Confident Reasoning (2025.findings-emnlp)

Copied to clipboard

Hyosoon Jang, Yunhui Jang, Sungjae Lee, Jungseul Ok, Sungsoo Ahn

Open-RAG: Enhanced Retrieval Augmented Reasoning with Open-Source Large Language Models (2024.findings-emnlp)

Challenge:	Large language models generate reasoning paths before final answers, but learning such a path requires costly human supervision.
Approach:	They propose a method that fine-tunes LLMs to prefer reasoning paths with high confidence . they propose 'cORE-PO' that fine tunes Lms to choose high-quality reasoning paths .
Outcome:	The proposed method improves the accuracy of outputs on four in-distribution and two out-of-difference benchmarks.

Copied to clipboard

Shayekh Islam, Md Asib Rahman, K S M Tozammel Hossain, Enamul Hoque, Shafiq Joty, Md Rizwan Parvez

Debating for Better Reasoning in Vision-Language Models (2025.findings-emnlp)

Challenge:	Existing methods to integrate Large Language Models with external knowledge suffer from limited reasoning capabilities, especially when using open-source LLMs.
Approach:	They propose a framework that transforms an arbitrary dense LLM into a parameter-efficient sparse mixture of experts (MoE) model capable of handling complex reasoning tasks.
Outcome:	The proposed framework transforms an arbitrary dense LLM into a parameter-efficient sparse mixture of experts (MoE) model capable of handling complex reasoning tasks, including both single- and multi-hop queries.

Copied to clipboard

Ashutosh Adhikari, Mirella Lapata

IP-Dialog: Evaluating Implicit Personalization in Dialogue Systems with Synthetic Data (2025.findings-emnlp)

Challenge:	Large Language Models (LLMs) gain expertise across diverse domains and modalities, a new study shows . scalable oversight becomes challenging when their capabilities surpass human evaluators.
Approach:	a new study extends the debate paradigm to a multimodal setting . it explores the potential for blind models to supervise and enhance the performance of sighted ones.
Outcome:	The proposed framework outperforms individual LLMs on multimodal tasks . it allows blind models to supervise and enhance the performance of sighted models .

Copied to clipboard

Bo Peng, Zhiheng Wang, Heyang Gong, Chaochao Lu

Natural Language Interfaces

Challenge:	Traditional data generation methods are labor-intensive, resource-demanding, and raise privacy concerns.
Approach:	They propose an automatic synthetic data generation approach and introduce the Implicit Personalized Dialogue benchmark along with a training dataset, covering 10 tasks and 12 user attribute types.
Outcome:	The proposed approach incorporates the Implicit Personalized Dialog**ue benchmark along with a training dataset, covering 10 tasks and 12 user attribute types.

Beyond Transcription: Unified Audio Schema for Perception-Aware AudioLLMs (2026.findings-acl)

Copied to clipboard

Linhao Zhang, Yuhan Song, Aiwei Liu, Chuhan Wu, Sijun Zhang, Wei Jia, Yuan Liu, Houfeng Wang, Zhou Xiao

V-ALPHASOCIAL: Benchmark and Self-Reflective Chain-of-Thought Generation for Visual Social Commonsense Reasoning (2025.findings-acl)

Challenge:	Recent Audio Large Language Models (AudioLLMs) excel at reasoning tasks, but struggle at elementary auditory perception.
Approach:	They propose a framework that organizes audio information into three explicit components in a unified JSON format.
Outcome:	The proposed framework boosts fine-grained perception by 10.9% on MMSU over state-of-the-art models while preserving robust reasoning capabilities.

Copied to clipboard

Zongyu Lin, Zhikun Xu, Xiaohan Song, Yixin Wan, Xingcheng Yao, Tsung-Han Lin, Selina Song, Pranav Subbaraman, Ben Zhou, Kai-Wei Chang, Yizhou Sun

Direct Behavior Optimization: Unlocking the Potential of Lightweight LLMs (2025.findings-acl)

Challenge:	Social commonsense reasoning is a multimodal task that requires both textual and visual cues.
Approach:	They propose a method that integrates visual cues into social commonsense reasoning tasks.
Outcome:	The proposed method improves social commonsense reasoning on a multimodal foundation model.

Copied to clipboard

Hongming Yang, Shi Lin, Jun Shao, Changting Lin, Donghai Zhu, Meng Han, Qinglei Kong

JX4MEI: Multimodal Semantically-Enhanced LLM for Joint Multimodal Emotion-Intent Explanation and Classification (2026.findings-acl)

Challenge:	Existing prompt optimization methods rely on extensive manual effort or meta-cognitive abilities, making them less effective for LwLLMs.
Approach:	They propose a direct behavior optimization parameter that transforms the optimization of complex prompts into discrete, quantifiable execution sequences using a gradient-free Monte Carlo Tree Search.
Outcome:	The proposed method outperforms current prompt optimization methods on seven challenging tasks where state-of-the-art LLMs excel but LwLLMs generally underperform.

Copied to clipboard

YiJie Huang, Xiaocui Yang, Shi Feng, Daling Wang, Yifei Zhang, Ning Yuan, Zhuoyue Jia, Wen Zhang

Information Retrieval

Challenge:	Existing multimodal emotion and intent recognition tasks focus on classification, not rationale and intrinsic connections between these states.
Approach:	They propose a task that requires models to jointly predict emotion and intent while generating natural language explanations for why they co-occur.
Outcome:	The proposed model outperforms baseline models in prediction and explanation generation.

Think Globally, Group Locally: Evaluating LLMs Using Multi-Lingual Word Grouping Games (2025.emnlp-main)

Copied to clipboard

César Guerra-Solano, Zhuochun Li, Xiang Lorraine Li

Reasoning-Aware AIGC Detection via Alignment and Reinforcement (2026.findings-acl)

Challenge:	Large language models exhibit biases in reasoning abilities due to linguistic modality, even with similar content.
Approach:	They propose a task inspired by the New York Times Connections: GlobalGroup that evaluates large language models in an abstract reasoning task across several languages.
Outcome:	The proposed task evaluates models across multiple languages in English, Spanish, Chinese, Hindi, and Arabic.

Copied to clipboard

Zhao Wang, Max Xiong, Jianxun Lian, Zhicheng Dou

VisuoThink: Empowering LVLM Reasoning with Multimodal Tree Search (2025.acl-long)

Challenge:	Existing approaches to AIGC detection have relied on statistical classifiers or black-box neural models, which exploit surface-level patterns and struggle to generalize as LLMs evolve.
Approach:	They propose a framework that generates interpretable reasoning chains before classification using supervised fine-tuning and reinforcement learning to improve accuracy.
Outcome:	The proposed framework achieves state-of-the-art performance across multiple benchmarks, offering a robust and transparent solution for AIGC detection.

Copied to clipboard

Yikun Wang, Siyin Wang, Qinyuan Cheng, Zhaoye Fei, Liang Ding, Qipeng Guo, Dacheng Tao, Xipeng Qiu

Verbosity-Aware Rationale Reduction: Sentence-Level Rationale Reduction for Efficient and Effective Reasoning (2025.findings-acl)

Challenge:	Existing approaches to large vision-language models fail to capture interleaved nature of human visual-verbal reasoning processes.
Approach:	They propose a framework that integrates visuospatial and linguistic domains to facilitate multimodal slow thinking by enabling progressive visual-textual reasoning.
Outcome:	Experiments show that VisuoThink significantly improves reasoning capabilities even without fine-tuning.

Copied to clipboard

Joonwon Jang, Jaehee Kim, Wonbin Kweon, Seonghyeon Lee, Hwanjo Yu

PLAN-TUNING: Post-Training Language Models to Learn Step-by-Step Planning for Complex Problem Solving (2025.emnlp-main)

Challenge:	Recent advances in Large Language Models (LLMs) demonstrate remarkable reasoning capabilities comparable to human cognitive abilities.
Approach:	They propose a sentence-level rationale reduction framework that leverages likelihood-based criteria to selectively remove redundant reasoning sentences while preserving reasoning capabilities.
Outcome:	The proposed framework improves performance by 7.71% while reducing token generation by 19.87% compared to models trained with complete reasoning paths.

Copied to clipboard

Mihir Parmar, Palash Goyal, Xin Liu, Yiwen Song, Mingyang Ling, Chitta Baral, Hamid Palangi, Tomas Pfister

CTRL: Control-Based Time Series Forecasting with LLM-Guided Residual Learning (2026.findings-acl)

Challenge:	Recent studies have shown that decomposing complex problems into simple subtasks has significantly boosted the performance of large language models (LLMs).
Approach:	They propose a unified post-training framework that distills synthetic task decompositions and fine-tunes smaller LLMs via supervised and reinforcement-learning objectives to improve complex reasoning.
Outcome:	The proposed framework outperforms strong baselines on GSM8k and MATH benchmarks and shows that it can improve generalization capabilities on out-of-domain datasets.

Copied to clipboard

Minkyoung Kim, Daeun Ji, Yohan Lee, Beomsoo Kim, Beakcheol Jang

Probe Then Retrieve and Reason: Distilling Probing and Reasoning Capabilities into Smaller Language Models (2024.lrec-main)

Challenge:	Existing time series forecasting approaches reduce them to numerical predictors that bypass their strengths or allow direct forecast generation that destabilizes predictions in non-stationary settings.
Approach:	They propose a framework that decouples semantic reasoning from quantitative prediction.
Outcome:	The proposed framework decouples semantic reasoning from quantitative prediction.

Copied to clipboard

Yichun Zhao, Shuheng Zhou, Huijia Zhu

Towards Stable and Effective Reinforcement Learning for Mixture-of-Experts (2026.acl-long)

Challenge:	Recent research efforts have focused on distilling Large Language Models into Small Language Model (SLMs) however, the results of CoT distillation are inadequate for knowledge-intensive reasoning tasks.
Approach:	They propose a retrieval-based framework which distills question probing and reasoning capabilities from Large Language Models into SLMs.
Outcome:	The proposed framework improves probing and reasoning capabilities of large language models in knowledge-intensive reasoning tasks.

Copied to clipboard

Di Zhang, Xun Wu, Shaohan Huang, Lingjie Jiang, Yaru Hao, Li Dong, Zewen Chi, Zhifang Sui, Furu Wei

MorphoBench: A Benchmark with Difficulty Adaptive to Model Reasoning (2026.findings-acl)

Challenge:	Reinforcement learning with verifiable rewards (RLVR) training with Mixture-of-Experts policies remains fragile and prone to reward collapse.
Approach:	They propose a router shift-based policy optimization method that computes a per-token router-shift ratio conditioned on the previously activated experts and applies stop-gradient and a lower-bound floor.
Outcome:	The proposed method achieves better performance and greater stability than previous methods.

Copied to clipboard

Xukai Wang, Xuanbo Liu, Mingrui Chen, Haitian Zhong, Xuanlin Yang, Bohan Zeng, Jinbo Hu, Hao Liang, Junbo Niu, Xuchen Li, Ruitao Wu, Ruichuan An, Yang Shi, Liu Liu, Qiang Liu, Zhouchen Lin, Xu-Yao Zhang, Wentao Zhang, Bin Dong

Language is Scary when Over-Analyzed: Unpacking Implied Misogynistic Reasoning with Argumentation Theory-Driven Prompts (2024.emnlp-main)

Challenge:	Existing benchmarks designed to evaluate the reasoning capabilities of large models are limited in scope and lack flexibility to adapt difficulty according to evolving reasoning capacities of models.
Approach:	They propose a benchmark that incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models.
Outcome:	The proposed benchmark incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models.

Copied to clipboard

Arianna Muti, Federico Ruggeri, Khalid Khatib, Alberto Barrón-Cedeño, Tommaso Caselli

An Empirical Analysis on Spatial Reasoning Capabilities of Large Multimodal Models (2024.emnlp-main)

Challenge:	a new study aims to understand the implicit reasoning used to convey misogynistic comments in Italian and English.
Approach:	They propose misogyny detection as an Argumentative Reasoning task and use argumentation theory to build large language models to understand the implicit reasoning used to convey misogany in Italian and English.
Outcome:	The proposed task is an argumentative reasoning task in Italian and English.

Copied to clipboard

Fatemeh Shiri, Xiao-Yu Guo, Mona Far, Xin Yu, Reza Haf, Yuan-Fang Li

Learning to Insert [PAUSE] Tokens for Better Reasoning (2025.findings-acl)

Challenge:	Large Multimodal Models (LMMs) have shown impressive generalization ability on vision and language tasks, but their spatial understanding is under-explored.
Approach:	They construct a VQA dataset to analyze LMMs' spatial reasoning capabilities.
Outcome:	The proposed model is stronger at basic object detection than complex spatial reasoning.

Copied to clipboard

Eunki Kim, Sangryul Kim, James Thorne

Reflection on Knowledge Graph for Large Language Models Reasoning (2025.findings-acl)

Challenge:	Existing studies have explored incorporating special-purpose tokens into the training process to enhance reasoning capabilities.
Approach:	They propose a method for inserting dummy tokens consecutively just before reasoning steps to increase model effectiveness.
Outcome:	The proposed method outperforms fine-tuning and previous token insertion methods on multiple datasets and models.

Copied to clipboard

Yigeng Zhou, Wu Li, Yifan Lu, Jing Li, Fangming Liu, Meishan Zhang, Yequan Wang, Daojing He, Honghai Liu, Min Zhang

Burn After Reading: Do Multimodal Large Language Models Truly Capture Order of Events in Image Sequences? (2025.findings-acl)

Challenge:	Existing methods for supplementing Large Language Models (LLMs) with knowledge graphs often introduce noise in the retrieval and reasoning pipeline, hindering their ability to integrate external knowledge for complex multi-hop question answering.
Approach:	They propose a framework to enhance LLMs' reasoning capabilities through reflective engagement with knowledge graphs by Query Decoupling, LLM-Driven Knowledge Graph Exploration, and Inference with Knowledge Reconstruction.
Outcome:	The proposed framework integrates external knowledge into LLMs and trains them to leverage this knowledge for answering questions.

Copied to clipboard

Yingjin Song, Yupei Du, Denis Paperno, Albert Gatt

KARPA: A Training-free Method of Adapting Knowledge Graph as References for Large Language Model’s Reasoning Path Aggregation (2025.findings-acl)

Challenge:	Existing benchmarks focus on single image settings, but some focus on multi-image settings.
Approach:	They introduce the TempVS benchmark which focuses on temporal grounding and reasoning capabilities of Multimodal Large Language Models in image sequences.
Outcome:	The proposed model performs poorly compared to human models in vision and language tasks.

Copied to clipboard

Siyuan Fang, Kaijing Ma, Tianyu Zheng, Xeron Du, Ningxuan Lu, Ge Zhang, Qingkun Tang

VCORE: Variance-Controlled Optimization-based Reweighting for Chain-of-Thought Supervision (2026.acl-long)

Challenge:	Existing methods for large language models (LLMs) are limited by step-by-step decision-making on KGs, or require fine-tuning or pre-training on specific KG.
Approach:	They propose a framework that harnesses the global planning abilities of large language models (LLMs) for efficient and accurate KG reasoning.
Outcome:	Extensive experiments show that the proposed framework achieves state-of-the-art performance in KGQA tasks, delivering both high efficiency and accuracy.

Copied to clipboard

Xuan Gong, Senmiao Wang, Hanbo Huang, Ruoyu Sun, Shiyu Liang

Summary Factual Inconsistency Detection Based on LLMs Enhanced by Universal Information Extraction (2025.findings-acl)

Challenge:	Empirical evaluations demonstrate that VCORE achieves the strongest overall average performance, with especially clear gains on lower-capacity models.
Approach:	They propose a framework that reformulates supervision as a constrained optimization problem.
Outcome:	Empirical evaluations show that VCORE achieves the strongest overall average performance, with especially clear gains on lower-capacity models.

Copied to clipboard

Anguo Li, Lei Yu

Information Extraction & Text Mining

Challenge:	Recent studies have shown that Large language models can detect factual inconsistencies in summaries but they lack the efficiency and explainability needed to be effective.
Approach:	They propose to decouple LLMs’ information extraction and reasoning capabilities to address key challenges and propose a framework for UIEFID to guide fine-tuned LLM methods in extracting unified structured information from documents and summaries.
Outcome:	The proposed framework improves the detection accuracy and reduces redundant reasoning on the AGGREFACT benchmark.

Temporal Sampling for Forgotten Reasoning in LLMs (2026.acl-long)

Copied to clipboard

Yuetai Li, Zhangchen Xu, Fengqing Jiang, Bhaskar Ramasubramanian, Luyao Niu, Bill Yuchen Lin, Xiang Yue, Radha Poovendran

RG-VQA: Leveraging Retriever-Generator Pipelines for Knowledge Intensive Visual Question Answering (2025.findings-emnlp)

Challenge:	a new metric measures the percentage of questions that were answered incorrectly during fine-tuning .
Approach:	They propose a decoding strategy that draws outputs from multiple checkpoints along the training trajectory.
Outcome:	The proposed method improves reasoning performance and consistency across benchmarks.

Copied to clipboard

Settaluri Lakshmi Sravanthi, Pulkit Agarwal, Debjyoti Mondal, Rituraj Singh, Subhadarshi Panda, Ankit Mishra, Kiran Pradeep, Srihari K B, Godawari Sudhakar Rao, Pushpak Bhattacharyya

Exposing the Achilles’ Heel: Evaluating LLMs Ability to Handle Mistakes in Mathematical Reasoning (2025.acl-long)

Challenge:	Existing methods to improve the reasoning capabilities of VQA systems are limited due to complexity of graph neural networks and end-to-end training.
Approach:	They propose a method to integrate Dense Passage Retrievers with Vision Language Models to boost the reasoning capabilities of VQA systems.
Outcome:	The proposed method outperforms human accuracy and GPT-4 in the ScienceQA dataset.

Copied to clipboard

Joykirat Singh, Akshay Nambi, Vibhav Vineet

Beyond Monolithic Rewards: Hybrid Multi-Aspect Reward Optimization (2026.findings-acl)

Challenge:	Existing evaluations focus on final accuracy, neglecting the critical aspect of reasoning capabilities.
Approach:	They propose to evaluate LLMs’ abilities to detect and correct reasoning mistakes by using rule-based methods and smaller language models.
Outcome:	The proposed model outperforms existing models such as GPT-4o and GPT4 in both accuracy and accuracy, but lacks data contamination and memorization concerns.

Copied to clipboard

Radha Gulhane, Sathish Reddy Indurthi

Hidden States as Early Signals: Step-level Trace Evaluation and Pruning for Efficient Test-Time Scaling (2026.findings-acl)

Challenge:	Existing approaches to optimize for multimodal learning use a single reward mechanism, but they lack confidence calibration across domains.
Approach:	They propose a hybrid reward and multi-aspect reward modeling framework that integrates model-based and rule-based reward paradigms for accuracy and confidence calibration.
Outcome:	The proposed model improves accuracy and confidence calibration across multimodal tasks and introduces a generalized length-penalty reward to stabilize training and improve performance.

Copied to clipboard

Zhixiang Liang, Beichen Huang, Zheng Wang, Minjia Zhang

What If Consensus Lies? Selective-Complementary Reinforcement Learning at Test Time (2026.acl-long)

Challenge:	Existing approaches to speed up parallel scaling have relied on similarity-based or confidence-based pruning, but these signals do not reliably indicate trace quality.
Approach:	They propose a pruning framework that evaluates reasoning steps using hidden states and dynamically prunes unpromising traces during generation.
Outcome:	The proposed framework reduces end-to-end inference latency by 45%–70% on average compared to self-consistency while improving reasoning accuracy.

Copied to clipboard

Dong Yan, Jian Liang, Yanbo Wang, Shuo Lu, Ran He, Tieniu Tan

GRAT: Guiding Retrieval-Augmented Reasoning through Process Rewards Tree Search (2025.acl-long)

Challenge:	Existing TTRL methods rely on positive pseudo-labeling strategies to enhance reasoning capabilities.
Approach:	They propose a test-time reinforcement learning framework that mitigates label noise amplification by deriving pseudo-rewards from majority voting consensus.
Outcome:	The proposed framework mitigates label noise amplification by implementing selective positive pseudo-labeling and entropy-gated negative p-labeled pruning.

Copied to clipboard

Xianshu Peng, Wei Wei

Information Retrieval

Challenge:	Existing methods to enhance large models for multi-hop question-answering lack the ability for multipath exploration, strategic look-ahead, stepwise evaluation, and global selection.
Approach:	They propose an algorithm guided by Monte Carlo Tree Search and process rewards that assigns fine-grained rewards to each step in the search path.
Outcome:	The proposed algorithm outperforms various RAG-based methods on four multihop QA datasets and shows that it can self-train and self-update.

CoVerRL: Breaking the Consensus Trap in Label-Free Reasoning via Generator-Verifier Co-Evolution (2026.acl-long)

Copied to clipboard

Teng Pan, Yuchen Yan, Zixuan Wang, Ruiqing Zhang, Guiyang Hou, Wenqi Zhang, Weiming Lu, Jun Xiao, Yongliang Shen

Beyond Correctness: Confidence-Aware Reward Modeling for Enhancing Large Language Model Reasoning (2025.emnlp-main)

Challenge:	Label-free reinforcement learning enables large language models to improve reasoning capabilities . but as training maximizes self-consistency, output diversity collapses, authors say . authors propose a framework where a single model alternates between generator and verifier roles .
Approach:	They propose a framework where a model alternates between generator and verifier roles, bootstrapping each other.
Outcome:	Experiments show that CoVerRL outperforms label-free baselines on reasoning benchmarks . the framework can be used to improve reasoning abilities without ground-truth supervision .

Copied to clipboard

Qianxi He, Qingyu Ren, Shanzhe Lei, Xuhong Wang, Yingchun Wang

We Argue to Agree: Towards Personality-Driven Argumentation-Based Negotiation Dialogue Systems for Tourism (2025.findings-emnlp)

Challenge:	Recent advances in large language models have shifted the post-training paradigm from instruction tuning and human preference alignment to reinforcement learning (RL) based on rule-based evaluations of answer correctness, these models often receive rewards for speculative answers without generating coherent reasoning chains.
Approach:	They propose a confidence-based reward model tailored for enhancing STEM reasoning capabilities.
Outcome:	The proposed model outperforms state-of-the-art open-source reward models across diverse STEM benchmarks.

Copied to clipboard

Priyanshu Priya, Saurav Dudhate, Desai Vishesh Yasheshbhai, Asif Ekbal

Dialogue Systems & Conversational Agents

Challenge:	Argumentation mechanisms are integrated into negotiation dialogue systems to improve conflict resolution and adaptability.
Approach:	They propose a dataset of Argumentation Profile, Preference Profile, and Buying Style Profiles to generate personality-driven dialogues in negotiation dialogue systems.
Outcome:	The proposed task improves argumentation mechanisms and adaptability by aligning interactions with individuals’ preferences and styles.

InquireMobile: Teaching VLM-based Mobile Agent to Request Human Assistance via Reinforcement Fine-Tuning (2026.acl-long)

Copied to clipboard

Qihang Ai, Pi Bu, Yue Cao, Yingyao Wang, Jihao Gu, Jingxuan Xing, Zekun Zhu, Wei Jiang, Zhicheng Zheng, Jun Song, Yuning Jiang

RouterHGC: Optimized Router for LLM-based Multi-Agent Systems via Heterogeneous Graph Contrastive Learning (2026.findings-acl)

Challenge:	Recent advances in Vision-Language Models (VLMs) have enabled mobile agents to perceive and interact with real-world mobile environments based on human instructions.
Approach:	They propose a vision-language model that actively seeks human confirmation at critical decision points and a model inspired by reinforcement learning.
Outcome:	The proposed model achieves an improvement of 46.8% in inquiry success rate and the best overall success rate among existing baselines on InquireBench.

Copied to clipboard

Yitao Xiao, Shaoyong Guo, Guoming Yang, Qingnan Wang, Yinlin Ren, Xuesong Qiu, Qi Feng

Turning Failures into Value: Negative Experience Replay for RLVR via Confidence Gating and Boundary Failure Sampling (2026.acl-long)

Challenge:	Large Language Models (LLMs)-driven Multi-Agent Systems (MAS) have demonstrated remarkable scalability and generalizability across complex tasks.
Approach:	They propose a new framework for routing using large language models . they formalize routing as node selection through edge-weight prediction .
Outcome:	The proposed framework outperforms the best single LLM and baselines on five datasets . it achieves 0.80%–6.17% accuracy gains on MATH and HotpotQA while reducing inference cost by 27.40%.

Copied to clipboard

Jialiang Guo, Fucheng Xiong, Xu He, Haodong Zhao, Xingyang li, Ke Zeng, Xunliang Cai

Taking Notes Brings Focus? Towards Multi-Turn Multimodal Dialogue Learning (2025.emnlp-main)

Challenge:	Existing experience replay methods for RLVR ignore sample inefficiency . expensive reasoning trajectories are discarded immediately after a single gradient update .
Approach:	They propose a method to replay failure trajectories to improve model refinement . they propose 'nexGRPO' which employs mid-confidence gating to filter invalid noise and saturated errors.
Outcome:	The proposed model outperforms strong baaselines and achieves improved out-of-distribution generalization.

Copied to clipboard

Jiazheng Liu, Sipeng Zheng, Börje F. Karlsson, Zongqing Lu

Natural Language Interfaces

Challenge:	Existing multimodal large language models are trained on single-turn vision question-answering tasks, which do not accurately reflect real-world human conversations.
Approach:	They propose a large-scale multi-turn multimodal dialogue dataset that uses rules and GPT assistance to generate a multi-turned multimodal dialog dataset.
Outcome:	The proposed dataset is a strong benchmark for multi-turn multimodal dialogue learning . it features complex dialogues with contextual dependencies that force models to track, ground, and recall information across multiple turns and disparate visual regions.

KRETA: A Benchmark for Korean Reading and Reasoning in Text-Rich VQA Attuned to Diverse Visual Contexts (2025.emnlp-main)

Copied to clipboard

Taebaek Hwang, Minseo Kim, Gisang Lee, Seonuk Kim, Hyunjun Eun

JoPR: Joint Emotion Perception and Reasoning for Conversational Emotion Recognition (2026.acl-long)

Challenge:	Understanding and reasoning over text within visual contexts poses a significant challenge for Vision-Language Models.
Approach:	They propose a benchmark for Korean Reading and rEasoning in Text-rich VQA Attuned to diverse visual contexts to address this challenge.
Outcome:	The proposed benchmark is tailored for Korean reading and rEasoning in text-rich VQA attuned to diverse visual contexts.

Copied to clipboard

Yumeng Fu, Weitao Huang, Junjie Wu, Hao Teng, Meishan Zhang, Bingquan Liu

Dialogue Systems & Conversational Agents

Challenge:	Existing methods for ERC lack human-like emotion reasoning and discrimination between similar emotions.
Approach:	They propose a multi-dimension curriculum with long CoT fine-tuning to clone human-like emotion reasoning for conversational emotion recognition.
Outcome:	The proposed model outperforms existing methods on three widely used datasets and shows that it is more intuitive and more accurate.

Sparse-RL: Breaking the Memory Wall in LLM Reinforcement Learning via Stable Sparse Rollouts (2026.acl-long)

Copied to clipboard

Sijia Luo, Xiaokang Zhang, Yuxuan Hu, Bohan Zhang, Ke Wang, Jinbo Su, Mengshu Sun, Lei Liang, Jing Zhang

Challenge:	Existing methods for storing key-value caches during long-horizon rollouts cause performance collapses.
Approach:	They propose a new training paradigm that empowers stable RL training under sparse rollouts.
Outcome:	The proposed model reduces rollout overhead while maintaining the performance.

rSIM: Incentivizing Reasoning Capabilities of LLMs via Reinforced Strategy Injection (2026.acl-long)

Copied to clipboard

Sijia Chen, Baochun Li, Di Niu

Benchmarking Fine-Grained Error Detection in Multimodal Reasoning (2026.acl-long)

Challenge:	Existing literature on Reasoning Language Models (RLMs) focuses on the ability to integrate reasoning strategies into the chain-of-thought process, contributing to improved problem-solving accuracy.
Approach:	They propose a reinforced strategy injection mechanism that enables any LLM to become an RLM by employing a small planner to guide the LLM's CoT through the adaptive injection of reasoning strategies.
Outcome:	The proposed model outperforms existing models in mathematical, coding, and financial reasoning tasks and is generalizable.

Copied to clipboard

Chi-Min Chan, Han Zhu, Chunyang Jiang, Jiaming Ji, Juntao Dai, Wei Xue, Sirui Han, Yike Guo

TRAC: Teacher-Guided Token Reward with Adaptive Calibration for Robust Policy Optimization (2026.acl-long)

Challenge:	Multimodal Process Reward Models (MPRMs) have emerged as a pivotal framework for enhancing the reasoning capabilities of Multimodal Large Language Models.
Approach:	They propose a benchmark specifically designed to evaluate MPRMs’ proficiency in detecting erroneous reasoning steps across diverse error categories.
Outcome:	The proposed model achieves up to 4.8% performance improvement through test-time scaling.

Copied to clipboard

Sitong Wu, Haoru Tan, Xichen Zhang, Bin Xia, Wenhu Zhang, Xiaojuan Qi, Bei Yu, Jiaya Jia