PruneCD: Contrasting Pruned Self Model to Improve Decoding Factuality
Comments accepted at EMNLP 2025 Main Conference
期刊&会议
EMNLP
Conference on Empirical Methods in Natural Language Processing · 会议 · Natural Language Processing
PolBiX: Detecting LLMs' Political Bias in Fact-Checking through X-phemisms
机构 * Quality & Usability Lab, Technische Universität Berlin（柏林技术大学质量与可用性实验室） ； German Research Center for Artificial Intelligence (DFKI), Berlin（德国人工智能研究中心（DFKI）） ； Department of Media and Communication, Ludwig-Maximilians-Universität München（慕尼黑路德维希-马克西米利安大学媒体与传播系）
Comments Accepted at Findings of EMNLP 2025, camera-ready version
Toxicity Red-Teaming: Benchmarking LLM Safety in Singapore's Low-Resource Languages
机构 * Singapore University of Technology and Design（新加坡科技设计大学） ； Mohamed bin Zayed University of Artificial Intelligence（马尔代夫 bin Zayed 人工智能大学）
Comments 9 pages, EMNLP 2025
Linguistic Neuron Overlap Patterns to Facilitate Cross-lingual Transfer on Low-resource Languages
机构 * School of Information Science and Technology, Beijing Foreign Studies University（信息科学与技术学院，北京外国语大学） ； Department of Informatics, King’s College London（信息学院，伦敦国王学院）
Comments Accepted by EMNLP 2025
Identifying and Answering Questions with False Assumptions: An Interpretable Approach
机构 * Department of Computer Science University of Arizona（计算机科学系亚利桑那大学）
Comments To appear at EMNLP 2025 Main conference
Automating Steering for Safe Multimodal Large Language Models
机构 * Zhejiang University（浙江大学） ； Zhejiang University - Ant Group Joint Lab of Knowledge Graph（浙江大学-蚂蚁集团知识图谱联合实验室） ； National University of Singapore, NUS-NCS Joint Lab（新加坡国立大学NUS-NCS联合实验室）
Comments EMNLP 2025 Main Conference. 23 pages (8+ for main); 25 figures; 1 table
LaMP-Cap: Personalized Figure Caption Generation With Multimodal Figure Profiles
机构 * The Pennsylvania State University（宾夕法尼亚州立大学） ； Adobe Research（Adobe研究）
Comments Accepted to EMNLP 2025 Findings. The LaMP-CAP dataset is publicly available at: https://github.com/Crowd-AI-Lab/lamp-cap
Please Translate Again: Two Simple Experiments on Whether Human-Like Reasoning Helps Translation
机构 * Language Technology Lab University of Amsterdam（语言技术实验室 阿姆斯特丹大学）
Comments EMNLP Main 2025 17 pages, 15 figures
Unraveling Misinformation Propagation in LLM Reasoning
机构 * EPFL（瑞士联邦理工学院） ； Stony Brook University（史泰士布鲁克大学） ； University of Chicago（芝加哥大学）
Comments Accepted to EMNLP 2025 (Findings)
NileChat: Towards Linguistically Diverse and Culturally Aware LLMs for Local Communities
机构 * The University of British Columbia（不列颠哥伦比亚大学） ； Mohammed VI Polytechnic University（穆莱·阿卜杜勒阿齐兹国王理工学院） ； Tuwaiq Academy（图瓦伊克学院） ； Invertible AI（可逆人工智能）
Comments Accepted to EMNLP 2025 (Main Conference). Camera-ready version. Data & models: https://github.com/UBC-NLP/nilechat
Memorization or Reasoning? Exploring the Idiom Understanding of LLMs
Comments EMNLP 2025
JOLT-SQL: Joint Loss Tuning of Text-to-SQL with Confusion-aware Noisy Schema Sampling
机构 * Zhengzhou University（郑州大学） ； Zhengzhou University of Aeronautics（郑州航空工业管理大学） ； Wuhan University（武汉大学）
Comments Accepted to EMNLP 2025 Main Conference
Mitigating Hallucination in Large Vision-Language Models through Aligning Attention Distribution to Information Flow
机构 * School of Computer Science and Technology, Beijing Institute of Technology（计算机科学与技术学院，北京理工大学） ； Zhongguancun Academy（中关村学院） ； Southeast Academy of Information Technology, Beijing Institute of Technology（信息技术东南学院，北京理工大学）
Comments Accepted to Findings of EMNLP 2025
Disambiguation in Conversational Question Answering in the Era of LLMs and Agents: A Survey
机构 * Adobe Research（Adobe研究部） ； KAIST（韩国科学技术院） ； Adobe Inc.（Adobe公司）
Comments 14 pages, 2 figures, Accepted at EMNLP 2025 Main Conference
Promote, Suppress, Iterate: How Language Models Answer One-to-Many Factual Queries
机构 * University of Southern California（南加州大学）
Comments Accepted to EMNLP 2025
Biology-Instructions: A Dataset and Benchmark for Multi-Omics Sequence Understanding Capability of Large Language Models
机构 * Shanghai Artificial Intelligence Laboratory（上海人工智能实验室） ； University of Science and Technology of China（中国科学技术大学） ； University of Sydney（悉尼大学） ； University of Toronto（多伦多大学） ； Chinese University of Hong Kong（香港中文大学） ； Shanghai Jiao Tong University（上海交通大学） ； Fudan University（复旦大学） ； Shanghai Innovation Institute（上海创新研究院）
Comments EMNLP 2025 findings
Post-hoc Study of Climate Microtargeting on Social Media Ads with LLMs: Thematic Insights and Fairness Evaluation
Comments Accepted at Findings of 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP 2025)
LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via a Hybrid Architecture
机构 * The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳）） ； Lehigh University（莱斯利大学） ； Meituan（美团）
Comments Accepted to EMNLP 2025 Findings
GraDeT-HTR: A Resource-Efficient Bengali Handwritten Text Recognition System utilizing Grapheme-based Tokenizer and Decoder-only Transformer
机构 * Computer Science and Engineering, University of Dhaka（计算机科学与工程系，达卡大学）
Comments 7 pages. Accepted at the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP) System Demonstrations. Equal Contribution: Md. Mahmudul Hasan and Ahmed Nesar Tahsin Choudhury
Improving Large Language Models Function Calling and Interpretability via Guided-Structured Templates
机构 * University of Notre Dame（诺丁汉大学） ； Amazon（亚马逊）
Comments Accepted to EMNLP 2025 Main Conference
RadEval: A framework for radiology text evaluation
机构 * University of Oxford（牛津大学） ； University of Glasgow（格拉斯哥大学） ； HOPPR
Comments Accepted to EMNLP 2025 Demo track - Oral
ReDepress: A Cognitive Framework for Detecting Depression Relapse from Social Media
机构 * Indian Institute of Technology Bombay（印度理工学院班加罗尔）
Comments Accepted to EMNLP 2025 Main Conference
Bringing Pedagogy into Focus: Evaluating Virtual Teaching Assistants' Question-Answering in Asynchronous Learning Environments
机构 * Columbia University（哥伦比亚大学）
Comments Accepted in EMNLP 2025 Findings
HICode: Hierarchical Inductive Coding with LLMs
Comments Long paper accepted at EMNLP 2025 main conference, 19 pages, 8 figures
How Persuasive is Your Context?
机构 * ETH Zürich（苏黎世联邦理工学院）
Comments Long paper accepted at EMNLP 2025
Make Every Letter Count: Building Dialect Variation Dictionaries from Monolingual Corpora
机构 * MaiNLP, Ludwig Maximilian University of Munich（MaiNLP，慕尼黑大学） ； Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心）
Comments Accepted at EMNLP 2025 (Findings)
Measuring Scalar Constructs in Social Science with LLMs
机构 * University of Innsbruck（因斯布鲁克大学） ； University of Maryland（马里兰大学） ； American University（美国大学） ； Northeastern University（东北大学） ； ETH Zürich（苏黎世联邦理工学院）
Comments Accepted to EMNLP 2025 (Main)
DCR: Quantifying Data Contamination in LLMs Evaluation
机构 * University College Dublin（都柏林大学学院） ； Georgia Institute of Technology（佐治亚理工学院）
Comments EMNLP 2025 Main
Med-PRM: Medical Reasoning Models with Stepwise, Guideline-verified Process Rewards
机构 * Korea University（韩国大学） ； ETH Zürich（苏黎世联邦理工学院） ； Yale University（耶鲁大学） ； AIGEN Sciences（AIGEN公司） ； Hanyang University College of Medicine（翰雅大学医学院） ； University of Ulsan College of Medicine（釜山大学医学院）
Comments Accepted to EMNLP 2025 (Oral)
PAKTON: A Multi-Agent Framework for Question Answering in Long Legal Agreements
机构 * School of Electrical and Computer Engineering, AILS Laboratory, National Technical University of Athens（电气与计算机工程学院，AILS实验室，国家技术大学雅典） ； Instituto de Telecomunicações（电信研究所）
Comments Accepted at EMNLP 2025
