EMNLP
Collection
Accepted papers for EMNLP (Conference on Empirical Methods in Natural Language Processing), one dataset per year. • 14 items • Updated
paper_id stringlengths 17 20 | title stringlengths 26 148 | paper_url stringlengths 43 46 | authors listlengths 1 60 | abstract large_stringlengths 494 2.51k | anthology_id stringlengths 17 20 | doi stringclasses 0
values | award stringclasses 0
values | arxiv_id stringlengths 10 10 ⌀ | arxiv_id_source stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|
2026.emnlp-main.1 | DeepSEA: A Multilingual Evaluation Benchmark for Southeast Asia | https://aclanthology.org/2026.emnlp-main.1/ | [
"Jiuzhou Han",
"Abhishek Bijay Mishra",
"Trevor Cohn",
"Shubham Mittal"
] | While language models have demonstrated remarkable capabilities, their performance remains disproportionately skewed towards high-resource languages. This disparity is particularly acute for the 700 million inhabitants of Southeast Asia, where vast linguistic diversity remains under-represented in existing benchmarks, ... | 2026.emnlp-main.1 | null | null | null | null |
2026.emnlp-main.2 | PopResume: Causal Fairness Evaluation of LLM/VLM Resume Screeners with Population-Representative Dataset | https://aclanthology.org/2026.emnlp-main.2/ | [
"Sumin Yu",
"Juhyeon Park",
"Taesup Moon"
] | We present PopResume, a population-representative resume dataset for causal fairness auditing of LLM- and VLM-based resume screening systems. Unlike existing benchmarks that rely on manually injected demographic information and outcome-level disparities, PopResume is grounded in population statistics and preserves natu... | 2026.emnlp-main.2 | null | null | 2603.22714 | title_snapshot |
2026.emnlp-main.3 | Logic Meets Language: Hybrid Query Answering over Incomplete Text-Labeled Graphs | https://aclanthology.org/2026.emnlp-main.3/ | [
"Yunjie He",
"Bo Xiong",
"Daniel Hernández",
"Yuqicheng Zhu",
"Yi Wang",
"Evgeny Kharlamov",
"Steffen Staab"
] | Text-labeled graph databases combine symbolic graph structures with rich textual descriptions, enabling expressive representations for real-world data. However, answering queries over such databases remains challenging when queries require both precise logical constraints and flexible semantic text interpretation, espe... | 2026.emnlp-main.3 | null | null | null | null |
2026.emnlp-main.4 | From Self to Other: Evaluating Demographic Perspective-Taking in LLM Hate Speech Annotation | https://aclanthology.org/2026.emnlp-main.4/ | [
"Paloma Piot",
"Javier Parapar"
] | Hate speech detection is inherently subjective: people from different demographic groups perceive the same content very differently. Collecting enough annotations from multiple demographic groups is costly and difficult to scale. Persona-conditioned Large Language Models (models prompted to adopt a specific demographic... | 2026.emnlp-main.4 | null | null | 2606.06266 | title_snapshot |
2026.emnlp-main.5 | The Hidden Toll of Social Media News on Psychosocial Wellbeing | https://aclanthology.org/2026.emnlp-main.5/ | [
"Olivia Pal",
"Agam Goyal",
"Eshwar Chandrasekharan",
"Koustuv Saha"
] | News consumption on social media has become ubiquitous, yet how different forms of engagement shape psychosocial outcomes remains unclear. To address this gap, we leveraged a large-scale dataset of ~26M posts and ~45M comments on the Bluesky platform and conducted a quasi-experimental study, matching 81,345 Treated use... | 2026.emnlp-main.5 | null | null | 2601.13487 | title_judge |
2026.emnlp-main.6 | FloCA: Towards Faithful and Logically Consistent Flowchart Reasoning | https://aclanthology.org/2026.emnlp-main.6/ | [
"Jinzi Zou",
"Bolin Wang",
"Shuo Zhang",
"Nuo Xu",
"Junzhou Zhao"
] | Flowchart-oriented dialogue (FOD) systems aim to guide users through multi-turn decision-making or operational procedures by following a domain-specific flowchart to achieve a task goal. In this work, we formalize flowchart reasoning in FOD as grounding user input to flowchart nodes at each dialogue turn while ensuring... | 2026.emnlp-main.6 | null | null | 2602.14035 | title_snapshot |
2026.emnlp-main.7 | The Trilemma of Truth in Large Language Models | https://aclanthology.org/2026.emnlp-main.7/ | [
"Germans Savcisens",
"Tina Eliassi-Rad"
] | The public often attributes human-like qualities to large language models (LLMs), assuming that they "know" certain things. In reality, LLMs encode information retained during training as internal probabilistic knowledge. This study examines existing methods for probing the veracity of that knowledge and identifies thr... | 2026.emnlp-main.7 | null | null | 2506.23921 | title_snapshot |
2026.emnlp-main.8 | Functional Entropy: Predicting Functional Correctness in LLM-Generated Code with Uncertainty Quantification | https://aclanthology.org/2026.emnlp-main.8/ | [
"Dylan Bouchard",
"Mohit Singh Chauhan",
"Zeya Ahmad",
"Ho-Kyeong Ra"
] | Large language models have shown impressive capabilities in code generation, yet they often produce functionally incorrect code. Uncertainty quantification (UQ) methods have emerged as a promising approach for detecting hallucinations in natural language generation, but their effectiveness for code generation tasks rem... | 2026.emnlp-main.8 | null | null | 2605.28500 | title_snapshot |
2026.emnlp-main.9 | Benchmarking Bengali Dialectal Bias: A Multi-Stage Framework Integrating RAG-Based Translation and Human-Augmented RLAIF | https://aclanthology.org/2026.emnlp-main.9/ | [
"K. M. Jubair Sami",
"Dipto Sumit",
"Ariyan Hossain",
"Farig Sadeque"
] | Large language models (LLMs) frequently exhibit performance biases against regional dialects of low-resource languages. However, frameworks to quantify these disparities remain scarce. We propose a two-phase framework to evaluate dialectal bias, operationalized as comprehension degradation relative to standard Bengali,... | 2026.emnlp-main.9 | null | null | 2603.21359 | title_snapshot |
2026.emnlp-main.10 | CALM: Class-Conditional Sparse Attention Vectors for Large Audio-Language Models | https://aclanthology.org/2026.emnlp-main.10/ | [
"Videet Mehta",
"Liming Wang",
"Hilde Kuehne",
"Rogerio Feris",
"James R. Glass",
"Muhammad Jehanzeb Mirza"
] | Large audio-language models (LALMs) show strong zero-shot performance on tasks such as audio question answering, but still lag behind specialized systems on discriminative tasks such as audio classification. Recent work shows that sparse subsets of attention heads can serve as discriminative feature extractors for few-... | 2026.emnlp-main.10 | null | null | 2602.07077 | title_snapshot |
2026.emnlp-main.11 | Check The Scoreboard: An Analysis of Scoring Schemes on Multiple-Choice Evaluation | https://aclanthology.org/2026.emnlp-main.11/ | [
"Nishant Balepur",
"Paiheng Xu",
"Wei Ai",
"Eunsol Choi",
"Rachel Rudinger",
"Jordan Boyd-Graber"
] | Multiple-choice question answering (MCQA) benchmarks in NLP use number-right scoring (accuracy), but in educational testing, the scoring scheme—the combination of the response mode models follow and the rule for grading responses—is a key design choice that dictates which abilities to reward. We examine how alternative... | 2026.emnlp-main.11 | null | null | 2608.29887 | title_snapshot |
2026.emnlp-main.12 | Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks | https://aclanthology.org/2026.emnlp-main.12/ | [
"Haoyu Liu",
"Dingcheng Li",
"Lukas Rutishauser",
"Zeyu Zheng"
] | Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet their dual-stream architecture opens an underexplored attack surface: an adversary who injects content into the webpage DOM simultaneously corrupts both observation channels with a ... | 2026.emnlp-main.12 | null | null | 2603.04364 | title_snapshot |
2026.emnlp-main.13 | Learning Personalized Prompts for Healthcare Guidance | https://aclanthology.org/2026.emnlp-main.13/ | [
"Ruize Shi",
"Hong Huang",
"Wei Zhou",
"Kehan Yin",
"Kai Zhao",
"Yun Zhao"
] | The rapid development of large language models (LLMs) has transformed many industries, including healthcare. In practice, hospitals and patients increasingly seek LLM-based systems capable of interpreting personal health records and providing healthcare guidance. However, existing approaches mainly rely on general medi... | 2026.emnlp-main.13 | null | null | 2412.15957 | title_snapshot |
2026.emnlp-main.14 | Expert Personas Improve LLM Alignment but Damage Accuracy: Bootstrapping Intent-Based Persona Routing with PRISM | https://aclanthology.org/2026.emnlp-main.14/ | [
"Zizhao Hu",
"Mohammad Rostami",
"Jesse Thomason"
] | Persona prompting can steer LLM generation towards a domain-specific tone and pattern. This behavior enables use cases in multi-agent systems where diverse interactions are crucial and human-centered tasks require high-level human alignment. Prior works provide mixed options on their utility: some report performance ga... | 2026.emnlp-main.14 | null | null | 2603.18507 | title_snapshot |
2026.emnlp-main.15 | Dementia-R1: Reinforced Pretraining and Reasoning from Unstructured Clinical Notes for Real-World Dementia Prognosis | https://aclanthology.org/2026.emnlp-main.15/ | [
"Choonghan Kim",
"Hyunmin Hwang",
"Hangeol Chang",
"Jaemin Kim",
"Jinse Park",
"Jae-Sung Lim",
"Jong Chul Ye"
] | While Large Language Models (LLMs) have shown strong performance on clinical text understanding, they struggle with longitudinal prediction tasks such as dementia prognosis, which require reasoning over complex, non-monotonic symptom trajectories across multiple visits. Standard supervised training lacks explicit annot... | 2026.emnlp-main.15 | null | null | 2601.03018 | title_snapshot |
2026.emnlp-main.16 | Granuscore: A Reference-Free Measure of Granularity for Text Analysis and Question Answering | https://aclanthology.org/2026.emnlp-main.16/ | [
"Lukas Ellinger",
"Alexander M. Fichtl",
"Miriam Anschütz",
"Georg Groh"
] | Natural language conveys information at varying levels of granularity, from fine-grained references to broad descriptions. While granularity is fundamental to human communication, existing measures mostly capture surface detail or sentence specificity. We introduce Granuscore, a reference-free measure of granularity th... | 2026.emnlp-main.16 | null | null | 2605.26620 | title_snapshot |
2026.emnlp-main.17 | Survival-Guided Length Control for Efficient Diffusion Language Models | https://aclanthology.org/2026.emnlp-main.17/ | [
"Ivan Kobyzev",
"Abbas Ghaddar",
"Yufei Cui"
] | Diffusion language models (DLMs) generate text by iteratively denoising masked sequences, but standard decoding either fixes the sequence length or relies on ad hoc stopping rules, often leading to unnecessary denoising steps. We recast length selection as a discrete-time survival problem over the end-of-sequence token... | 2026.emnlp-main.17 | null | null | 2608.26374 | title_snapshot |
2026.emnlp-main.18 | Measuring the Depth of LLM Unlearning via Activation Patching | https://aclanthology.org/2026.emnlp-main.18/ | [
"Jaeung Lee",
"Dohyun Kim",
"Jaemin Jo"
] | Large language model (LLM) unlearning has emerged as a crucial post-hoc mechanism for privacy protection and AI safety, yet auditing whether target knowledge is truly erased remains challenging. Existing output-level metrics fail to detect when this knowledge remains recoverable from internal representations. Recent wh... | 2026.emnlp-main.18 | null | null | 2605.24614 | title_snapshot |
2026.emnlp-main.19 | Compressing Sequences in the Latent Embedding Space: K-Token Merging for Large Language Models | https://aclanthology.org/2026.emnlp-main.19/ | [
"Zihao Xu",
"John Harvill",
"Ziwei Fan",
"Yizhou Sun",
"Hao Ding",
"Hao Wang"
] | Large Language Models (LLMs) incur significant computational and memory costs when processing long prompts, as full self-attention scales quadratically with input length. Token compression aims to address this challenge by reducing the number of tokens representing inputs. However, existing prompt-compression approache... | 2026.emnlp-main.19 | null | null | 2604.15153 | title_snapshot |
2026.emnlp-main.20 | AI Psychosis: Does Conversational AI Amplify Delusion-Related Language? | https://aclanthology.org/2026.emnlp-main.20/ | [
"Soorya Ram Shimgekar",
"Vipin Gunda",
"Jenny Jiwon Kim",
"Violeta J. Rodriguez",
"Hari Sundaram",
"Koustuv Saha"
] | Conversational AI systems are increasingly used for personal reflection and emotional disclosure, raising concerns about their effects on vulnerable users. Recent anecdotal reports suggest that prolonged interactions with conversational AI may reinforce delusional thinking—a phenomenon sometimes described as AI Psychos... | 2026.emnlp-main.20 | null | null | 2603.19574 | title_snapshot |
2026.emnlp-main.21 | Dissecting Transformers: A CLEAR Perspective Towards Green AI | https://aclanthology.org/2026.emnlp-main.21/ | [
"Hemang Jain",
"Shailender Goyal",
"Divyansh Pandey",
"Karthik Vaidhyanathan"
] | The rapid adoption of Large Language Models (LLMs) has raised significant environmental concerns. Unlike the one-time cost of training, LLM inference occurs continuously and dominates the AI energy footprint. Yet most sustainability studies report only coarse model-level metrics, treating energy efficiency as an aftert... | 2026.emnlp-main.21 | null | null | 2510.02810 | title_snapshot |
2026.emnlp-main.22 | Learning Diagnostic Clarifying Questions for Conversational Question Answering via Data-Efficient Dynamic Distillation | https://aclanthology.org/2026.emnlp-main.22/ | [
"Jihyeong Jeon",
"Cheol Ryu",
"U Kang"
] | Given a query, how can a QA system ask an informative clarifying question to accurately answer it? Open-domain conversational question answering requires not only detecting when a query is ambiguous but also asking clarifying questions that efficiently uncover the latent intent of the user. However, existing approaches... | 2026.emnlp-main.22 | null | null | null | null |
2026.emnlp-main.23 | PolySQL: Scaling Text-to-SQL Evaluation Across SQL Dialects via Automated Backend Isomorphism | https://aclanthology.org/2026.emnlp-main.23/ | [
"Yotam Perlitz",
"Elad Venezian",
"Corentin Royer",
"Francesco Fusco",
"Andrea Giovannini"
] | SQL dialects vary in syntax, types, and functions across database engines. Text-to-SQL benchmarks, however, predominantly support only SQLite. This creates a critical evaluation gap: cross-dialect evaluation reveals weak per-query agreement (Cohen’s ), showing that SQLite performance is an unreliable proxy for other di... | 2026.emnlp-main.23 | null | null | 2605.07796 | title_snapshot |
2026.emnlp-main.24 | Improving Diffusion Language Model Decoding through Joint Search in Generation Order and Token Space | https://aclanthology.org/2026.emnlp-main.24/ | [
"Yangyi Shen",
"Tianjian Feng",
"Jiaqi Han",
"Wen Wang",
"Tianlang Chen",
"Chunhua Shen",
"Jure Leskovec",
"Stefano Ermon"
] | Diffusion Language Models (DLMs) offer order-agnostic generation that can explore many possible decoding trajectories. However, current decoding methods commit to a single trajectory, limiting exploration in trajectory space. We introduce Order-Token Search to explore this space through jointly searching over generatio... | 2026.emnlp-main.24 | null | null | 2601.20339 | title_snapshot |
2026.emnlp-main.25 | PsyAgent: Constructing Human-like Agents Based on Psychological Modeling and Contextual Interaction | https://aclanthology.org/2026.emnlp-main.25/ | [
"Zibin Meng",
"Kani Chen"
] | Human-like agents must express stable dispositions while adapting to roles, relationships, and norms. We present PsyAgent, a schema-first framework that operationalizes the trait–context interface by coupling a Big Five trait prior with explicit social-structural conditioning. PsyAgent comprises (i) Individual Structur... | 2026.emnlp-main.25 | null | null | 2601.06158 | title_snapshot |
2026.emnlp-main.26 | CRU-Reason: Knowledge Graph-Grounded and Interpretable Clinical Prediction with Clinical Rationale Units | https://aclanthology.org/2026.emnlp-main.26/ | [
"Zihang Liu",
"Linhao Luo",
"Zhangchi Qiu",
"Shigeng Chen",
"Haishuai Wang"
] | Healthcare prediction using electronic health records (EHRs) is a cornerstone of modern clinical decision support. Conventional machine learning (ML) models achieve strong predictive performance but rarely provide knowledge-grounded, human-readable explanations. Large language models (LLMs) incorporated with medical kn... | 2026.emnlp-main.26 | null | null | null | null |
2026.emnlp-main.27 | From Text to Structure: The PACKaGE Benchmark for Bidirectional Causal Reasoning in Paragraphs and Graphs | https://aclanthology.org/2026.emnlp-main.27/ | [
"Juhyeon Kim",
"Dohun Lee",
"Jiwon Jang",
"Joseph Shim",
"Hyunwoo Park"
] | Large Language Models (LLMs) have demonstrated remarkable capabilities across various domains, yet they struggle with causal reasoning tasks. We introduce PACKaGE (Paragraph-level Annotated Causal Knowledge and Graph–tExt), a causal benchmark dataset that pairs paragraph-level texts with explicit causal graphs. Unlike ... | 2026.emnlp-main.27 | null | null | null | null |
2026.emnlp-main.28 | Efficient Iterative Retrieval with Heterogeneous Batching | https://aclanthology.org/2026.emnlp-main.28/ | [
"Dohyun Park",
"Hubertus Franke",
"Daniel G Waddington",
"Swaminathan Sundararaman",
"Yongjoo Park"
] | Modern information retrieval increasingly employs both embedding and generative models to handle complex queries. However, current serving systems suffer from low throughput and poor GPU utilization because they execute these models in isolation. Coarse-grained partitioning, such as dedicating GPUs to specific tasks, f... | 2026.emnlp-main.28 | null | null | 2609.25405 | title_snapshot |
2026.emnlp-main.29 | Spexis: Speculative Lookahead Scheduling for LLM Inference | https://aclanthology.org/2026.emnlp-main.29/ | [
"Hyungyu Jung",
"Jaehyeok Yu",
"Hoonseo Choi",
"Sungkyun Kim",
"Jinho Lee",
"Jiwon Seo"
] | Spexis is a multi-GPU LLM inference framework that improves the efficiency of pipeline and tensor parallelism through speculative parallelism. Rather than using speculative decoding only to accelerate token generation, Spexis runs speculation in parallel with normal execution, introducing a new parallelism axis without... | 2026.emnlp-main.29 | null | null | 2609.34370 | title_snapshot |
2026.emnlp-main.30 | When Does Auxiliary Classification Help Regression in Practice? A Case Study in Arabic Dialectness Modeling | https://aclanthology.org/2026.emnlp-main.30/ | [
"Abdelrahman Sakr",
"Marwan Torki",
"Nagwa El-Makky"
] | Estimating the Arabic Level of Dialectness (ALDi) is a continuous regression task that quantifies how much a sentence diverges from Modern Standard Arabic (MSA). While auxiliary classification is often used to support regression, its effectiveness depends critically on how the auxiliary task is designed. We conduct a s... | 2026.emnlp-main.30 | null | null | null | null |
2026.emnlp-main.31 | Cross-Attention Eigenvalues Know the Word Order in Multilingual Machine Translation | https://aclanthology.org/2026.emnlp-main.31/ | [
"Nikolai Prudnikov",
"Oleg Kachan"
] | Multilingual machine translation models must adapt to the grammatical structure of the target language, yet how structural information shapes decoder computation remains unclear. Using the eigenvalue spectra of per-(layer, head) attention Gram matrices instead of raw attention weights, we show that cross-attention geom... | 2026.emnlp-main.31 | null | null | null | null |
2026.emnlp-main.32 | Gold vs. Translation: Are We Measuring Language, Knowledge, or Locality in Translated Benchmarks? | https://aclanthology.org/2026.emnlp-main.32/ | [
"Klaudia-Doris Thellmann",
"Bernhard Stadler",
"Dan Saattrup Smart",
"Michael Färber",
"Jens Lehmann"
] | Translated versions of English-origin benchmarks such as MMLU are widely used to evaluate large language models in non-English languages, yet it remains unclear whether they measure the same constructs as natively created benchmarks. We frame this as a *proxy validity* question and compare DeepL-translated MMLU proxies... | 2026.emnlp-main.32 | null | null | null | null |
2026.emnlp-main.33 | From Passive Response to Proactive Correction: Enhancing LLM Robustness Against Input Fact Perturbations | https://aclanthology.org/2026.emnlp-main.33/ | [
"Ping Wang",
"Xiangguo Sun",
"Bingbing Xu",
"Guocong Li",
"Xiaofeng Meng"
] | Large language models (LLMs) frequently produce confident yet factually incorrect responses when user inputs contain misleading premises, a phenomenon we attribute to fact perturbations in the input. Existing approaches to hallucination mitigation typically assume reliable user inputs, overlooking how such factual erro... | 2026.emnlp-main.33 | null | null | 2608.25894 | title_snapshot |
2026.emnlp-main.34 | How Human-Like Are Large Language Models? A Register-Aware Linguistic Evaluation Framework | https://aclanthology.org/2026.emnlp-main.34/ | [
"Björn Nieth",
"Marianna Gracheva",
"Michaela Mahlberg",
"Bjoern Eskofier",
"Emmanuelle Salin"
] | While factual correctness and task-performance have received much attention by Large Language Model (LLM) research, the fundamental question of how human-like generated texts are on a linguistic level has been underexplored. From a corpus-linguistic perspective, language production is inherently context-dependent. Dist... | 2026.emnlp-main.34 | null | null | 2605.23651 | title_snapshot |
2026.emnlp-main.35 | Ontology-Driven Structural Regularization for Document-Level Relation Extraction | https://aclanthology.org/2026.emnlp-main.35/ | [
"Laura Menotti",
"Stefano Marchesin",
"Gianmaria Silvello"
] | Document-Level Relation Extraction (DocRE) relies heavily on costly manually annotated datasets, while large distant supervision resources such as DocRED distant remain underexploited due to noise. We show that a critical yet overlooked source of noise lies in structural inconsistencies within relational triples, inclu... | 2026.emnlp-main.35 | null | null | 2608.20856 | title_snapshot |
2026.emnlp-main.36 | SCALE: Structured Case-law-guided Anticipatory Legal Reasoning | https://aclanthology.org/2026.emnlp-main.36/ | [
"Mann Khatri",
"Santosh T.Y.S.S",
"Oana Ichim",
"Rajiv Ratn Shah",
"Ponnurangam Kumaraguru"
] | LLMs have shown promise in legal NLP, but generating legal reasoning that is factually grounded, faithful to precedent, and consistent with judicial analysis remains difficult. Retrieval-augmented generation is often used to improve grounding, yet standard legal RAG faces two limitations: retrieved materials are hard t... | 2026.emnlp-main.36 | null | null | null | null |
2026.emnlp-main.37 | Reasoning Under Uncertainty: Exploring Probabilistic Reasoning Capabilities of LLMs | https://aclanthology.org/2026.emnlp-main.37/ | [
"Mobina Pournemat",
"Keivan Rezaei",
"Gaurang Sriramanan",
"Arman Zarei",
"Hamid Eghbalzadeh",
"Yang Wang",
"Jiaxiang Fu",
"Fuchun Peng",
"Zhigang Wang",
"Xue Feng",
"Nima Noorshams",
"Soheil Feizi"
] | Large language models (LLMs) have demonstrated strong performance across a wide range of reasoning tasks, yet their understanding of joint and conditional probability distributions remains largely unexplored. In this work, we present a comprehensive evaluation of LLMs’ ability to reason about discrete joint and conditi... | 2026.emnlp-main.37 | null | null | 2509.10739 | title_snapshot |
2026.emnlp-main.38 | E^3-Pruner: Towards Efficient, Economical, and Effective Layer Pruning for Large Language Models | https://aclanthology.org/2026.emnlp-main.38/ | [
"Tao Yuan",
"Haoli Bai",
"Pan Yinfei",
"Xuyang Cao",
"Tianyu Zhang",
"Lu Hou",
"Ting Hu",
"Xianzhi Yu"
] | With the increasing size of large language models, layer pruning has gained increased attention as a hardware-friendly approach for model compression. However, existing layer pruning methods struggle to simultaneously address key practical deployment challenges, including performance degradation, high training costs, a... | 2026.emnlp-main.38 | null | null | 2511.17205 | title_snapshot |
2026.emnlp-main.39 | Forget What’s Sensitive, Remember What Matters: Token-Level Differential Privacy in Memory Sculpting for Continual Learning | https://aclanthology.org/2026.emnlp-main.39/ | [
"Bihao Zhan",
"Jie Zhou",
"Junsong Li",
"Yutao Yang",
"Shilian Chen",
"Qianjun Pan",
"Xin Li",
"Wen Wu",
"Xingjiao Wu",
"Qin Chen",
"Hang Yan",
"Liang Dou",
"Liang He"
] | Continual Learning (CL) models, while adept at sequential knowledge acquisition, face significant and often overlooked privacy challenges due to accumulating diverse information. Traditional privacy methods, like a uniform Differential Privacy (DP) budget, indiscriminately protect all data, leading to substantial model... | 2026.emnlp-main.39 | null | null | 2509.12958 | title_snapshot |
2026.emnlp-main.40 | Learning to Zoom Efficiently with a Contrastive Curriculum | https://aclanthology.org/2026.emnlp-main.40/ | [
"Falko Helm",
"Iryna Gurevych"
] | Using a zoom-in tool is an important foundational part of modern visual agents, because it allows to efficiently handle tasks involving high-resolution images. Most previous methods need an extensive warm-start supervised fine-tuning phase for teaching models zoom-in. We show that this is not necessary by proposing a n... | 2026.emnlp-main.40 | null | null | 2609.03206 | title_snapshot |
2026.emnlp-main.41 | From Leaky Thoughts to Private Reasoning: Controlling What LRMs Say to Themselves | https://aclanthology.org/2026.emnlp-main.41/ | [
"Haritz Puerto",
"Haonan Li",
"Xudong Han",
"Timothy Baldwin",
"Iryna Gurevych"
] | Large reasoning models (LRMs) produce reasoning traces (RTs) that often contain sensitive information. These leaky thoughts are difficult to control and frequently violate explicit privacy directives. Because RTs can be exposed through prompt injection attacks, this becomes a direct privacy risk to the user. We approac... | 2026.emnlp-main.41 | null | null | 2602.24210 | title_snapshot |
2026.emnlp-main.42 | OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis | https://aclanthology.org/2026.emnlp-main.42/ | [
"Zhuofeng Li",
"Dongfu Jiang",
"Xueguang Ma",
"Haoxiang Zhang",
"Ping Nie",
"Yuyu Zhang",
"Kai Zou",
"Jianwen Xie",
"Yu Zhang",
"Wenhu Chen"
] | Training deep research agents requires long-horizon trajectories that interleave search, evidence aggregation, and multi-step reasoning. However, existing data collection pipelines typically rely on proprietary web APIs, making large-scale trajectory synthesis costly, unstable, and difficult to reproduce. We present Op... | 2026.emnlp-main.42 | null | null | 2603.20278 | title_snapshot |
2026.emnlp-main.43 | Do Implicit Personalization and Explicit Styles Conflict? PsPLUG: A Lightweight Plug-in for Balancing Personalization and Style in Customized LLMs | https://aclanthology.org/2026.emnlp-main.43/ | [
"Yutong Song",
"Jiang Wu",
"Shaofan Yuan",
"Chengze Shen",
"Jian Wang",
"Yu Wang",
"Nikil Dutt",
"Amir M. Rahmani"
] | Personalized large language models are often expected to follow explicit style instructions, yet we find that such instructions can undermine the user-specific characteristics that personalization methods aim to preserve. We call this failure mode personalization collapse: explicit style control can conflict with impli... | 2026.emnlp-main.43 | null | null | 2601.06362 | title_snapshot |
2026.emnlp-main.44 | Retrieving Facts or Reinforcing Beliefs? Towards Understanding Confirmation Bias in RAG Systems for Scientific Claim Verification | https://aclanthology.org/2026.emnlp-main.44/ | [
"Qile Wan",
"Davide Venditti",
"Fabio Massimo Zanzotto",
"Iryna Gurevych",
"Lizhen Qu"
] | The rapid growth of scientific literature has made automated verification of scientific claims increasingly important, with Retrieval-Augmented Generation (RAG) systems emerging as a promising solution. However, their multi-stage architecture raises concerns about confirmation bias, where systems preferentially retriev... | 2026.emnlp-main.44 | null | null | null | null |
2026.emnlp-main.45 | Client Simulation via Conversational Trajectory Modeling for Both Trainee Practice and Model Evaluation in Mental Health Counseling | https://aclanthology.org/2026.emnlp-main.45/ | [
"Huachuan Qiu",
"Zhaoming Chen",
"Yuqian Chen",
"Yuan Xie",
"Yu Lu",
"Zhenzhong Lan"
] | LLM-based client simulation provides a scalable approach to novice counselor training, counseling-dialogue synthesis, and interactive evaluation of automated counseling systems. However, existing approaches are limited by insufficient profile diversity, weak behavioral grounding, and the lack of open Chinese-language r... | 2026.emnlp-main.45 | null | null | 2601.07312 | title_judge |
2026.emnlp-main.46 | MEMORA: A Memory-Enhanced Multimodal Committee Reranking Agent for Reasoning-Intensive Retrieval | https://aclanthology.org/2026.emnlp-main.46/ | [
"Mohamed Mahmoud",
"Mostafa Farouk Senussi",
"Mahmoud Abdalla",
"Mahmoud SalahEldin Kasem",
"Abdelrahman Abdallah",
"Hyun Soo Kang"
] | Reasoning-intensive retrieval is fundamentally challenged when queries demand deep semantic inference and arrive in multimodal form. Existing systems treat each query in isolation, rely on monolithic rerankers, and lack retrievers trained for cross-modal reasoning. We introduce MEMORA, a Memory-Enhanced Multimodal Orch... | 2026.emnlp-main.46 | null | null | null | null |
2026.emnlp-main.47 | SpecSteer: Synergizing Local Context and Global Reasoning for Efficient Personalized Generation | https://aclanthology.org/2026.emnlp-main.47/ | [
"Hang Lv",
"Sheng Liang",
"Hao Wang",
"Yongyue Zhang",
"Hongchao Gu",
"Wei Guo",
"Defu Lian",
"Yong Liu",
"Enhong Chen"
] | Realizing personalized intelligence faces a core dilemma: sending user history to centralized large language models raises privacy concerns, while on-device small language models lack the reasoning capacity required for high-quality generation. Our pilot study shows that purely local enhancements remain insufficient to... | 2026.emnlp-main.47 | null | null | 2603.16219 | title_snapshot |
2026.emnlp-main.48 | Revisiting Backdoor Attacks on LLMs: A Stealthy and Practical Poisoning Framework via Harmless Inputs | https://aclanthology.org/2026.emnlp-main.48/ | [
"Jiawei Kong",
"Hao Fang",
"Xiaochen Yang",
"Kuofeng Gao",
"Bin Chen",
"Shu-Tao Xia",
"Yaowei Wang",
"Min Zhang"
] | Recent backdoor studies on Large Language Models (LLMs) typically insert harmful question-answer (QA) pairs into their training data. However, we identify two critical limitations: (1) directly embedding harmful content into training data compromises safety alignment, resulting in attack efficacy even for clean queries... | 2026.emnlp-main.48 | null | null | 2505.17601 | title_snapshot |
2026.emnlp-main.49 | Exposía: Teaching and Assessment of Academic Writing Skills for Research Project Proposals and Peer Feedback | https://aclanthology.org/2026.emnlp-main.49/ | [
"Dennis Zyska",
"Alla Rozovskaya",
"Ilia Kuznetsov",
"Iryna Gurevych"
] | We present Exposía, the first public dataset that connects writing and feedback in higher education, enabling research on educationally grounded computational approaches to teaching and evaluating academic writing. Exposía includes student research project proposals and peer and instructor feedback consisting of commen... | 2026.emnlp-main.49 | null | null | 2601.06536 | title_snapshot |
2026.emnlp-main.50 | GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuning | https://aclanthology.org/2026.emnlp-main.50/ | [
"Zhouxiang Fang",
"Jiawei Zhou",
"Hanjie Chen"
] | Recent studies show that the safety alignment of large language models (LLMs) can be easily compromised even by seemingly non-adversarial fine-tuning. To preserve safety alignment during fine-tuning, a widely used strategy is to jointly optimize safety and task objectives by mixing in the original alignment data, which... | 2026.emnlp-main.50 | null | null | 2603.10243 | title_snapshot |
2026.emnlp-main.51 | What Makes a Good Response? An Empirical Analysis of Quality in Qualitative Interviews | https://aclanthology.org/2026.emnlp-main.51/ | [
"Jonathan Ivey",
"Anjalie Field",
"Ziang Xiao"
] | Qualitative interviews provide essential insights into human experiences when they elicit high-quality responses. While qualitative and NLP researchers have proposed various measures of interview quality, these measures lack validation that high-scoring responses actually contribute to the study’s goals. In this work, ... | 2026.emnlp-main.51 | null | null | 2604.05163 | title_snapshot |
2026.emnlp-main.52 | The Autonomy Tax: Defense Training Breaks LLM Agents | https://aclanthology.org/2026.emnlp-main.52/ | [
"Li Li",
"Yue Zhao"
] | Large language model (LLM) agents increasingly rely on external tools (file operations, API calls, database transactions) to autonomously complete complex multi-step tasks. Practitioners deploy defense-trained models to protect against prompt injection attacks that manipulate agent behavior through malicious observatio... | 2026.emnlp-main.52 | null | null | 2603.19423 | title_snapshot |
2026.emnlp-main.53 | Multi-Expert Conformal Risk Control for Pairwise LLM Judging in Open-Ended Dialogue | https://aclanthology.org/2026.emnlp-main.53/ | [
"Ming Cheng",
"Yusheng Dai",
"Qiuhong Ke",
"Zhaolin Chen",
"Lizhen Qu"
] | In this paper, we explore multi-expert Conformal Risk Control (CRC) algorithms for pairwise LLM-as-a-Judge evaluation in open-ended dialogue. Our core insight is that multi-expert aggregation offers a complementary remedy to CRC: whereas CRC controls risk at the decision threshold through abstention, aggregation saniti... | 2026.emnlp-main.53 | null | null | 2608.26529 | title_snapshot |
2026.emnlp-main.54 | ReGround: Grounding Reviewer Comments in Multimodal Evidence | https://aclanthology.org/2026.emnlp-main.54/ | [
"Serwar Basch",
"Lizhen Qu",
"Iryna Gurevych"
] | Reviewer comments naturally relate to specific parts of the reviewed paper, yet grounding these comments to the underlying evidence is difficult due to long multimodal documents. Existing benchmarks do not capture this setting and largely focus on explicit, information-seeking queries. We introduce ReGround, a large-sc... | 2026.emnlp-main.54 | null | null | 2609.11460 | title_snapshot |
2026.emnlp-main.55 | Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity? | https://aclanthology.org/2026.emnlp-main.55/ | [
"Anmol Goel",
"Iryna Gurevych"
] | Computer-use agents (CUAs) now act on a user’s behalf across personal applications such as email, calendars, and to-do lists. This cross-application access is useful, but it also creates a privacy risk that has been largely overlooked: when an agent works in one context, it can pull in information from another that is ... | 2026.emnlp-main.55 | null | null | 2606.23189 | title_snapshot |
2026.emnlp-main.56 | PIVOTS-Bench: Evaluating Fine-Grained Interpersonal Relationship Reasoning in Multimodal Large Language Models | https://aclanthology.org/2026.emnlp-main.56/ | [
"Shuxiang Zhang",
"Ciossa Yin",
"Wenxuan Song",
"Yuhang Wu",
"Miao Liu"
] | Humans possess an innate ability to understand fine-grained interpersonal relationships, which is central to everyday social interactions. Although such reasoning is inherently multimodal, it remains largely unexplored by existing multimodal large language models (MLLMs). To address this gap, we introduce PIVOTS-Bench,... | 2026.emnlp-main.56 | null | null | 2606.23092 | title_judge |
2026.emnlp-main.57 | Retrieval-Augmented LLM Agents: Learning to Learn from Experience | https://aclanthology.org/2026.emnlp-main.57/ | [
"Thomas Palmeira Ferraz",
"Romain Deffayet",
"Vassilina Nikoulina",
"Hervé Déjean",
"Stéphane Clinchant"
] | While large language models (LLMs) have advanced the development of general-purpose agents, robust generalization to unseen tasks remains challenging. Two common approaches are supervised fine-tuning and training-free memory-augmented generation using retrieved experience; yet both have limitations: fine-tuning often f... | 2026.emnlp-main.57 | null | null | 2603.18272 | title_snapshot |
2026.emnlp-main.58 | Conformal Privacy Auditing: Calibrated Re-identification Attacks with Statistical Guarantees | https://aclanthology.org/2026.emnlp-main.58/ | [
"Shuo Huang",
"Gholamreza Haffari",
"Xingliang Yuan",
"Lizhen Qu"
] | Empirical identity leakage from released text is increasingly driven by attackers that combine large language models (LLMs) with auxiliary knowledge to link documents to individuals. Existing audits typically report success rates for specific attack pipelines but lack finite-sample statistical guarantees, while trainin... | 2026.emnlp-main.58 | null | null | 2609.21340 | title_snapshot |
2026.emnlp-main.59 | SocialCritic: Grounding Multi-Agent Dialogue via Advantage-Based Endorsement Suppression | https://aclanthology.org/2026.emnlp-main.59/ | [
"Shuang Cao",
"Rui Li"
] | In multi-agent dialogue, a hallucination crosses an operational control boundary when another agent accepts it and downstream decisions inherit the error. We formalize this cross-speaker transition as unsupported endorsement and introduce SocialCritic, a learned receding-horizon controller that ranks and repairs candid... | 2026.emnlp-main.59 | null | null | null | null |
2026.emnlp-main.60 | Demographic Prompting at Scale: When More Attributes Hurt LLM–Human Agreement | https://aclanthology.org/2026.emnlp-main.60/ | [
"Mahammed Kamruzzaman",
"Shrabon Kumar Das",
"Gene Louis Kim"
] | We investigate how annotator demographic attributes, supplied as prompt cues, shape the alignment between large language model (LLM) predictions and human annotations across five tasks. Using five open-source LLMs, we systematically vary the number and composition of demographic components in the prompt, spanning every... | 2026.emnlp-main.60 | null | null | 2607.10590 | title_snapshot |
2026.emnlp-main.61 | SGM: Safety Glasses for Multimodal Large Language Models via Neuron-Level Detoxification | https://aclanthology.org/2026.emnlp-main.61/ | [
"Hongbo Wang",
"AprilPyone MaungMaung",
"Isao Echizen"
] | Disclaimer: Samples in this paper may be harmful and cause discomfort.Multimodal large language models (MLLMs) enable multimodal understanding but inherit toxic signals from weakly curated pretraining corpora, leading to explicitly toxic outputs, especially under adversarial triggers that late, opaque training-free det... | 2026.emnlp-main.61 | null | null | 2512.15052 | title_snapshot |
2026.emnlp-main.62 | The Evaluation Scaling Law: Toward More Reliable LLM Assessment via Multi-Prompt Diversity | https://aclanthology.org/2026.emnlp-main.62/ | [
"Hongzhi Luan",
"Zhaoxin Huan",
"Linfeng Shi",
"Longfei Zheng",
"Xiaolu Zhang",
"Jun Zhou"
] | Large Language Model (LLM) evaluation is frequently challenged by prompt sensitivity, which limits the reliability of benchmarks relying on a single template. To address this, we systematically investigate a multi-prompt evaluation framework that aggregates performance across diverse templates, leveraging a Quality-Div... | 2026.emnlp-main.62 | null | null | null | null |
2026.emnlp-main.63 | ALEE: Any-Language Evaluation of Embeddings via English-Centric Minimal Pairs | https://aclanthology.org/2026.emnlp-main.63/ | [
"Andrianos Michail",
"Stylianos Psychias",
"Michelle Wastl",
"Simon Clematide",
"Rico Sennrich",
"Juri Opitz"
] | Text embeddings are standard for semantic similarity tasks, yet their evaluation remains an open challenge. Current benchmarks are static, cover only a limited set of languages, are often domain-specific, susceptible to overfitting, and poorly representative of low-resource languages. To address these limitations, we i... | 2026.emnlp-main.63 | null | null | 2607.00171 | title_snapshot |
2026.emnlp-main.64 | Decompose, Look, and Reason: Reinforced Latent Reasoning for VLMs | https://aclanthology.org/2026.emnlp-main.64/ | [
"Mengdan Zhu",
"Senhao Cheng",
"Liang Zhao"
] | Vision-Language Models often struggle with complex visual reasoning due to the visual information loss in textual CoT. Existing methods either add the cost of tool calls or rely on localized patch-based embeddings that are insufficient to extract semantics in multi-step reasoning. We propose "Decompose, Look, and Reaso... | 2026.emnlp-main.64 | null | null | 2604.07518 | title_snapshot |
2026.emnlp-main.65 | CoPRM: A Collaborative Process Reward Model for Guiding Search Agent Reasoning | https://aclanthology.org/2026.emnlp-main.65/ | [
"Xingquan Zhang",
"Yang Zhao",
"Chong Li",
"Donglei Yu",
"Pu Jian",
"Chengqing Zong"
] | Large Language Model (LLM)-based search agents have demonstrated superior performance in knowledge-intensive tasks through interleaved reasoning and search interactions. Process Reward Models (PRMs) can enhance search agent performance by guiding step selection during inference. However, existing PRMs for search agents... | 2026.emnlp-main.65 | null | null | null | null |
2026.emnlp-main.66 | Who Laughs with Whom? Disentangling Influential Factors in Humor Preferences across User Clusters and LLMs | https://aclanthology.org/2026.emnlp-main.66/ | [
"Soichiro Murakami",
"Hidetaka Kamigaito",
"Hiroya Takamura",
"Manabu Okumura"
] | Humor preferences vary widely across individuals and cultures, complicating the evaluation of humor using large language models (LLMs). In this study, we model heterogeneity in humor preferences in Oogiri, a Japanese creative response game, by clustering users with voting logs and estimating cluster-specific weights ov... | 2026.emnlp-main.66 | null | null | 2601.03103 | title_snapshot |
2026.emnlp-main.67 | ClaimFlow: Tracing the Evolution of Scientific Claims in NLP | https://aclanthology.org/2026.emnlp-main.67/ | [
"Aniket Pramanick",
"Yufang Hou",
"Saif M. Mohammad",
"Iryna Gurevych"
] | Scientific papers advance claims that later work supports, extends, or sometimes refutes. Yet existing methods for citation and claim analysis capture only fragments of this dialogue. In this work, we make these interactions explicit at the level of scientific claims. We introduce ClaimFlow, a claim-centric view of the... | 2026.emnlp-main.67 | null | null | 2603.16073 | title_snapshot |
2026.emnlp-main.68 | T-MAP: Red-Teaming LLM Agents with Trajectory-aware Evolutionary Search | https://aclanthology.org/2026.emnlp-main.68/ | [
"Hyomin Lee",
"Sangwoo Park",
"Yumin Choi",
"Sohyun An",
"Seanie Lee",
"Sung Ju Hwang"
] | While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models (LLMs), such approaches fail to capture agent-specific vulnerabilities that emerge through multi-step tool execution, particularly in rapidly growing ecosystems such as the Model Context Protocol (MCP). To address ... | 2026.emnlp-main.68 | null | null | 2603.22341 | title_snapshot |
2026.emnlp-main.69 | Mini-BEHAVIOR-Gran: Revealing U-Shaped Effects of Instruction Granularity on Language-Guided Embodied Agents | https://aclanthology.org/2026.emnlp-main.69/ | [
"Sukai Huang",
"Chenyuan Zhang",
"Fucai Ke",
"Zhixi Cai",
"Gholamreza Haffari",
"Lizhen Qu",
"Hamid Rezatofighi"
] | Instruction granularity is an important yet poorly controlled variable in language-guided embodied AI. Existing benchmarks typically pair each task with a single static instruction, making it difficult to study how agent behavior changes when the same task is described at different levels of detail. We introduce Mini-B... | 2026.emnlp-main.69 | null | null | 2604.17019 | title_snapshot |
2026.emnlp-main.70 | Diverging Transformer Predictions for Human Sentence Processing: A Comprehensive Analysis of Agreement Attraction Effects | https://aclanthology.org/2026.emnlp-main.70/ | [
"Titus von der Malsburg",
"Sebastian Padó"
] | Transformers underlie almost all state-of-the-art language models in computational linguistics, yet their cognitive adequacy as models of human sentence processing remains disputed. In this work, we use a surprisal-based linking mechanism to systematically evaluate eleven autoregressive transformers of varying sizes an... | 2026.emnlp-main.70 | null | null | 2603.16574 | title_snapshot |
2026.emnlp-main.71 | NanoVDR: Distilling a 2B Vision-Language Retriever into a 70M Text-Only Encoder for Visual Document Retrieval | https://aclanthology.org/2026.emnlp-main.71/ | [
"Zhuchenyang Liu",
"Yao Zhang",
"Yu Xiao"
] | Vision-Language Model (VLM) based retrievers have advanced visual document retrieval (VDR) to impressive quality. They require the same multi-billion parameter encoder for both document indexing and query encoding, incurring high latency and GPU dependence even for plain-text queries. We observe that this design is unn... | 2026.emnlp-main.71 | null | null | 2603.12824 | title_snapshot |
2026.emnlp-main.72 | Evaluating LLM-generated Explanatory Dialogue Turns through Dialogue Completion | https://aclanthology.org/2026.emnlp-main.72/ | [
"Yi-Sheng Hsu",
"Nils Feldhus",
"Manfred Stede"
] | Synthetic explanatory dialogues hold promise for applications from tutoring systems to model rationalization, yet their quality is often deemed inferior to human-authored conversations. This study uses dialogue completion as a lens to investigate the nuanced differences between human-spoken, human-written, and syntheti... | 2026.emnlp-main.72 | null | null | null | null |
2026.emnlp-main.73 | Who Tests the Testers? Systematic Enumeration and Coverage Audit of LLM Agent Tool Call Safety | https://aclanthology.org/2026.emnlp-main.73/ | [
"Xuan Chen",
"Lu Yan",
"Ruqi Zhang",
"Xiangyu Zhang"
] | Large Language Model (LLM) agents increasingly act through external tools, making their safety contingent on tool-call workflows rather than text generation alone. While recent benchmarks evaluate agents across diverse environments and risk categories, a fundamental question remains unanswered: how complete are existin... | 2026.emnlp-main.73 | null | null | 2603.18245 | title_snapshot |
2026.emnlp-main.74 | LogicEnvGen: Task-Logic Driven Generation of Diverse Simulated Environments for Embodied AI | https://aclanthology.org/2026.emnlp-main.74/ | [
"Jianan Wang",
"Siyang Zhang",
"Bin Li",
"Juan Chen",
"Jingtao Qi",
"Zhuo Zhang",
"Chen Qian"
] | Simulated environments play an essential role in embodied AI, functionally analogous to test cases in software engineering. However, existing environment generation methods often emphasize visual realism (e.g., object diversity and layout coherence), overlooking a crucial aspect: logical diversity from the testing pers... | 2026.emnlp-main.74 | null | null | 2601.13556 | title_snapshot |
2026.emnlp-main.75 | Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection | https://aclanthology.org/2026.emnlp-main.75/ | [
"Harethah Abu Shairah",
"Hasan Abed Al Kader Hammoud",
"George Turkiyyah",
"Bernard Ghanem"
] | Safety alignment in Large Language Models (LLMs) often involves mediating internal representations to refuse harmful requests. Recent research has demonstrated that these safety mechanisms can be bypassed by ablating or removing specific representational directions within the model. In this paper, we propose the opposi... | 2026.emnlp-main.75 | null | null | 2508.20766 | title_snapshot |
2026.emnlp-main.76 | Efficient LLM Safety Evaluation through Multi-Agent Debate | https://aclanthology.org/2026.emnlp-main.76/ | [
"Dachuan Lin",
"Guobin Shen",
"Zihao Yang",
"Tianrong Liu",
"Dongcheng Zhao",
"Yi Zeng"
] | Safety evaluation of large language models (LLMs) increasingly relies on LLM-as-a-judge pipelines, but strong judges can still be costly to deploy at scale. We present a multi-agent judge that uses small language models (SLMs) in structured debates among critic, defender, and judge agents. To evaluate these systems, we... | 2026.emnlp-main.76 | null | null | 2511.06396 | title_snapshot |
2026.emnlp-main.77 | TTSR: Test-Time Self-Evolving via Reflection | https://aclanthology.org/2026.emnlp-main.77/ | [
"Haoyang He",
"Zihua Rong",
"Yunjia Zhao",
"Lan Yang",
"Jian Chang",
"Honggang Zhang"
] | Test-time training (TTT) adapts large language models (LLMs) during inference using only unlabeled test inputs. Existing methods, however, face two major bottlenecks on hard reasoning tasks: (1) lack of learnable samples, as self-generated pseudo-labels on difficult questions are often noisy and yield unstable rewards;... | 2026.emnlp-main.77 | null | null | 2603.03297 | title_snapshot |
2026.emnlp-main.78 | Library Hallucinations in LLM-Generated Code: A Risk Analysis Grounded in Developer Queries | https://aclanthology.org/2026.emnlp-main.78/ | [
"Lukas Twist",
"Mark Harman",
"Helen Yannakoudakis",
"Jie M. Zhang"
] | Large language models (LLMs) now play a central role in code generation, yet they continue to hallucinate, frequently inventing non-existent libraries. Such library hallucinations are not just benign errors: they can mislead developers, break builds, and expose systems to supply chain threats such as slopsquatting. Des... | 2026.emnlp-main.78 | null | null | 2509.22202 | title_snapshot |
2026.emnlp-main.79 | What Drives Representation Steering? A Mechanistic Case Study on Steering Refusal | https://aclanthology.org/2026.emnlp-main.79/ | [
"Stephen S Cheng",
"Sarah Wiegreffe",
"Dinesh Manocha"
] | Applying steering vectors to large language models (LLMs) is an efficient and effective model alignment technique, but we lack an interpretable explanation for how it works–specifically, what internal mechanisms steering vectors affect and how this results in different model outputs. To investigate the causal mechanism... | 2026.emnlp-main.79 | null | null | 2604.08524 | title_snapshot |
2026.emnlp-main.80 | Calibrated Tree-Structured Speculative Decoding: Mitigating Overconfidence via Path-Level Calibration | https://aclanthology.org/2026.emnlp-main.80/ | [
"Zixian LI",
"Xiaohui Song",
"Chi Xie",
"Ke Chen",
"Chen Chen",
"Haonan Lu"
] | Tree-structured speculative decoding accelerates large language model inference by drafting multiple candidate tokens in parallel and verifying them against a target model. Dynamic tree expansion strategies further improve efficiency by using draft model confidence to prune unlikely branches and stop expansion early. H... | 2026.emnlp-main.80 | null | null | null | null |
2026.emnlp-main.81 | Fast and Focused: Interactive Repository-Level Fill-in-the-Middle Code Completion with Hybrid Structural Retrieval | https://aclanthology.org/2026.emnlp-main.81/ | [
"Long S. T. Nguyen",
"Dat Huynh",
"Quynh Vo",
"LE Dang Nguyen Hai",
"Cong-Duy T Nguyen",
"Anh Tuan Luu",
"Anh Nguyen-Duc",
"Tho Quan"
] | Large language models are effective at code completion when the required evidence is local, but repository-level fill-in-the-middle completion remains challenging when it depends on code and structural relationships across files. We present ACAR, a framework that combines lexical and AST/LSP-based structural retrieval ... | 2026.emnlp-main.81 | null | null | null | null |
2026.emnlp-main.82 | Preserving Historical Truth: Detecting Historical Revisionism in Large Language Models | https://aclanthology.org/2026.emnlp-main.82/ | [
"Francesco Ortu",
"Joeun Yook",
"Punya Syon Pandey",
"Bernhard Schölkopf",
"Alberto Cazzaniga",
"Rada Mihalcea",
"Zhijing Jin"
] | Large language models (LLMs) are increasingly consulted as sources of historical information, yet their handling of contested events and politically charged narratives remains understudied. We introduce HistoricalMisinfo, a dataset of 500 contested events from 45 countries, each paired with a factual and a documented r... | 2026.emnlp-main.82 | null | null | 2602.17433 | title_snapshot |
2026.emnlp-main.83 | Observable Patterns Are Not Explanations: A Causal-Geometric Analysis of Latent Reasoning Models | https://aclanthology.org/2026.emnlp-main.83/ | [
"Darpan Aswal",
"Thomas Palmeira Ferraz",
"Yongxin Zhou",
"Maxime Peyrard"
] | Latent reasoning models (LRMs) replace explicit chain-of-thought with continuous thoughts. Recent work treats observable latent-state patterns, such as BFS-like frontiers and decodable arithmetic computation, as evidence for internal reasoning mechanisms. Evaluating two LRMs (Coconut and CODI) against controls lacking ... | 2026.emnlp-main.83 | null | null | 2606.12689 | title_snapshot |
2026.emnlp-main.84 | Gated Residual Discrepancy for Efficient Bi-Encoder Factuality Verification | https://aclanthology.org/2026.emnlp-main.84/ | [
"Zeyang Liu",
"Jingfeng Xue",
"Jiarun Fu",
"Wenbiao Du",
"Xiuqi Yang",
"Yifeng Fu",
"Junbao Chen",
"Yong Wang"
] | Large Language Models (LLMs) often generate specious hallucinations, creating an urgent need for post-hoc factuality verification against reliable evidence. Existing verification methods incur substantial computational overhead due to joint evidence–claim encoding. While bi-encoders offer a highly efficient alternative... | 2026.emnlp-main.84 | null | null | null | null |
2026.emnlp-main.85 | Graph2Counsel: Clinically Grounded Synthetic Counseling Dialogue Generation from Client Psychological Graphs | https://aclanthology.org/2026.emnlp-main.85/ | [
"Aishik Mandal",
"Hiba Arnaout",
"Clarissa W. Ong",
"Juliet Bockhorst",
"Kate Sheehan",
"Rachael Moldow",
"Tanmoy Chakraborty",
"Iryna Gurevych"
] | Rising demand for mental health support has increased interest in using Large Language Models (LLMs) for counseling, but adapting them to this safety-critical domain is hindered by limited real-world data due to privacy constraints. Synthetic datasets provide a promising alternative, but existing approaches often rely ... | 2026.emnlp-main.85 | null | null | 2604.20382 | title_snapshot |
2026.emnlp-main.86 | LLM-Based Adversarial Persuasion Attacks on Fact-Checking Systems | https://aclanthology.org/2026.emnlp-main.86/ | [
"João A. Leite",
"Olesya Razuvayevskaya",
"Kalina Bontcheva",
"Carolina Scarton"
] | Automated fact-checking (AFC) systems are susceptible to adversarial attacks, enabling false claims to evade detection. Existing adversarial frameworks typically rely on injecting noise or altering semantics, yet no existing framework exploits the adversarial potential of persuasion techniques against AFC systems, whic... | 2026.emnlp-main.86 | null | null | 2601.16890 | title_snapshot |
2026.emnlp-main.87 | SpokenUS: A Spoken User Simulator for Task-Oriented Dialogue | https://aclanthology.org/2026.emnlp-main.87/ | [
"Jonggeun Lee",
"Junseong Pyo",
"Jeongmin Park",
"Yohan Jo"
] | Robust voice agents require exposure to the full diversity of how people interact through speech. However, obtaining enough spoken interactions is prohibitively expensive. Building spoken user simulators that address this requires large-scale spoken task-oriented dialogue (TOD) data encompassing spoken user behaviors, ... | 2026.emnlp-main.87 | null | null | 2603.16783 | title_snapshot |
2026.emnlp-main.88 | LENS: Exploiting the Cross-View Stability Gap for Hallucination Mitigation in Medical Vision–Language Models | https://aclanthology.org/2026.emnlp-main.88/ | [
"Dwarikanath Mahapatra",
"Abhijit Das",
"Sudipta Roy",
"Behzad Bozorgtabar",
"Imran Razzak"
] | Medical VLMs frequently hallucinate clinically unsupported findings. We identify a stability gap: visually grounded tokens maintain relatively stable logits under semantics-preserving image perturbations, whereas visually misgrounded tokens exhibit substantially larger shifts. We introduce LENS (Logit-ENhanced Stabilit... | 2026.emnlp-main.88 | null | null | null | null |
2026.emnlp-main.89 | CORE-T: COherent REtrieval of Tables for Text-to-SQL | https://aclanthology.org/2026.emnlp-main.89/ | [
"Hassan Soliman",
"Vivek Gupta",
"Dan Roth",
"Iryna Gurevych"
] | Realistic text-to-SQL workflows often require joining multiple tables. As a result, accurately retrieving the relevant set of tables becomes a key bottleneck for end-to-end performance. We study an open-book setting where queries must be answered over large, heterogeneous table collections pooled from many sources, wit... | 2026.emnlp-main.89 | null | null | 2601.13111 | title_snapshot |
2026.emnlp-main.90 | Unsupervised Speech Recognition at the Syllable Level | https://aclanthology.org/2026.emnlp-main.90/ | [
"Liming Wang",
"Kai-Wei Chang",
"Kunio Kashino",
"David Harwath",
"Mark A. Hasegawa-Johnson",
"James R. Glass"
] | Training speech recognizers with unpaired speech and text – known as unsupervised speech recognition (UASR) – is a crucial step toward extending ASR to low-resource languages in the long-tail distribution and enabling multimodal learning from non-parallel data. However, existing approaches based on phones often rely on... | 2026.emnlp-main.90 | null | null | 2608.22907 | title_snapshot |
2026.emnlp-main.91 | HybridThink: Learned Interleaving of Continuous and Discrete Reasoning in Large Language Models | https://aclanthology.org/2026.emnlp-main.91/ | [
"Jing Gao",
"Weiqiang Jin",
"Guoshuai Zhao",
"Peigen Ye",
"Xueming Qian",
"Hongyang Du"
] | Reasoning in Large Language Models (LLMs) is primarily conducted in a discrete token space. Although recent approaches employing continuous concept tokens facilitate abstract reasoning, they introduce significant stability challenges. The accumulation of Out-of-Distribution (OOD) states frequently causes manifold devia... | 2026.emnlp-main.91 | null | null | null | null |
2026.emnlp-main.92 | Entropic properties of languages in phonetic versus orthographic alphabets | https://aclanthology.org/2026.emnlp-main.92/ | [
"Carla Olabe",
"Christian Bentz"
] | Computational linguistics predominantly operates on written text. However, writing represents spoken language only indirectly. Whether phonetic and orthographic representations share fundamental information-theoretic properties is an open question. We present a cross-linguistic analysis of Shannon entropy across modali... | 2026.emnlp-main.92 | null | null | null | null |
2026.emnlp-main.93 | ViD: Vision-Dominant Gender Bias Mitigation for Large Vision-Language Models | https://aclanthology.org/2026.emnlp-main.93/ | [
"Zhipeng Zhao",
"Zhaoqiang Wei",
"Peishun Liu",
"Youwei Zhao",
"Ruichun Tang"
] | Gender bias in large vision-language models (LVLMs) undermines their fairness and reliability, compromising output trustworthiness. Current mitigation methods rely on training-phase adjustments or post-hoc calibration, but face limitations in dynamic visual bias mitigation. These include inability to capture real-time ... | 2026.emnlp-main.93 | null | null | 2609.16647 | title_snapshot |
2026.emnlp-main.94 | Learning to Detect Unseen Jailbreak Attacks in Large Vision-Language Models | https://aclanthology.org/2026.emnlp-main.94/ | [
"Shuang Liang",
"Zhihao Xu",
"Jiaqi Weng",
"Jialing Tao",
"Hui Xue",
"Xiting Wang"
] | Despite extensive alignment efforts, Large Vision-Language Models (LVLMs) remain vulnerable to jailbreak attacks. To mitigate these risks, existing detection methods are essential, yet they face two major challenges: generalization and accuracy. While learning-based methods trained on specific attacks fail to generaliz... | 2026.emnlp-main.94 | null | null | 2508.09201 | title_snapshot |
2026.emnlp-main.95 | Are Non-English Papers Reviewed Fairly? Language-of-Study Bias in NLP Peer Reviews | https://aclanthology.org/2026.emnlp-main.95/ | [
"Ehsan Barkhordar",
"Abdulfattah Safa",
"Verena Blaschke",
"Erika Lombart",
"Marie-Catherine de Marneffe",
"Gözde Gül Şahin"
] | Peer review plays a central role in the NLP publication process, but is susceptible to various biases. Here, we study language-of-study (LoS) bias: the tendency for reviewers to evaluate a paper differently based on the language(s) it studies, rather than its scientific merit. Despite being explicitly flagged in review... | 2026.emnlp-main.95 | null | null | 2604.07119 | title_snapshot |
2026.emnlp-main.96 | Self-Calibrating Language Models via Test-Time Discriminative Distillation | https://aclanthology.org/2026.emnlp-main.96/ | [
"Mohamed Rissal Hedna",
"Jan Strich",
"Martin Semmann",
"Chris Biemann"
] | Large language models (LLMs) are systematically overconfident: they routinely express high certainty on questions they often answer incorrectly. Existing calibration methods either require labeled validation data, degrade under distribution shifts, or incur substantial inference costs. Recent work has shown that LLMs a... | 2026.emnlp-main.96 | null | null | 2604.09624 | title_snapshot |
2026.emnlp-main.97 | VITAL: Visual-Semantic Dual Supervision for Enhanced and Interpretable Latent Reasoning in Medical MLLMs | https://aclanthology.org/2026.emnlp-main.97/ | [
"Qiaoru Li",
"Shaotian Liang",
"Jintao Chen",
"Haoran Sun",
"Yuxiang Cai",
"Jianwei Yin",
"Yankai Jiang"
] | Latent reasoning enables reasoning over continuous hidden states rather than explicit tokens, avoiding the language bottleneck and inference overhead of chain-of-thought for medical VQA. However, existing methods suffer from modality collapse, insufficient visual supervision, and train-inference mismatch. Moreover, the... | 2026.emnlp-main.97 | null | null | 2605.28422 | title_snapshot |
2026.emnlp-main.98 | Structure-Aware Data Augmentation for Argument Mining | https://aclanthology.org/2026.emnlp-main.98/ | [
"Masayuki Kawarada",
"Tsutomu Hirao",
"Manabu Okumura"
] | This paper introduces a novel data augmentation approach designed for argument mining. Argument mining is a task that automatically identifies argumentative structures within texts, typically represented as trees or directed acyclic graphs (DAGs). A major challenge facing current argument mining systems is the limited ... | 2026.emnlp-main.98 | null | null | null | null |
2026.emnlp-main.99 | From Reasoning to Generalization: Knowledge-Augmented LLMs for the ARC Benchmark | https://aclanthology.org/2026.emnlp-main.99/ | [
"Chao Lei",
"Krista A. Ehinger",
"Nir Lipovetzky",
"Yanchuan Chang"
] | Despite extensive research on reasoning-oriented LLMs, core cognitive faculties of human intelligence, such as abstract reasoning and generalization, remain underexplored. To address this, we evaluate recent reasoning-oriented LLMs on the Abstraction and Reasoning Corpus (ARC) benchmark, which explicitly demands both f... | 2026.emnlp-main.99 | null | null | 2505.17482 | title_judge |
2026.emnlp-main.100 | PRKV: Page Restructed KV Cache for High Accuracy and Efficiency LLM Generation | https://aclanthology.org/2026.emnlp-main.100/ | [
"Fang Wu",
"Congming Gao",
"Weixi Zhu",
"Jiwu Shu"
] | As the key-value(KV) cache size scales with context length, accessing large KV cache each step and substantial GPU memory demand challenge us to deploy LLMs with long contexts.Various sparse attention methods have been proposed and offloading-based KV retrieval preserves entire KV cache in CPU memory and dynamically re... | 2026.emnlp-main.100 | null | null | null | null |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 32