Papers by Wenya Wang

Unlocking LLMs’ Self-Improvement Capacity with Autonomous Learning for Domain Adaptation (2025.findings-acl)

Challenge:	Fact-checking pipelines adopt the Decompose-Then-Verify paradigm, where texts are broken down into smaller claims for individual verification and subsequently combined for a veracity decision.
Approach:	They propose to categorize decomposition errors and to reveal a trade-off between accuracy gains and noise introduced by decomposing.
Outcome:	The proposed analysis provides new insights into understanding current system’s instability and offers guidance for future studies toward improving claim decomposition in fact-checking pipelines.

Copied to clipboard

Ke Ji, Junying Chen, Anningzhe Gao, Wenya Xie, Xiang Wan, Benyou Wang

Elaboration-Generating Commonsense Question Answering at Scale (2023.acl-long)

Challenge:	Existing models that use self-supervised and instruction fine-tuning can be trained using unlabeled corpora.
Approach:	They propose to use unlabeled target corpora to adapt large language models to new domains . they propose to employ self-supervised pre-training and instruction fine-tuning methods .
Outcome:	The proposed model can adapt to new domains using only a large amount of unlabeled target corpora.

Copied to clipboard

Wenya Wang, Vivek Srikumar, Hannaneh Hajishirzi, Noah A. Smith

Recursive Neural Structural Correspondence Network for Cross-domain Aspect and Opinion Co-Extraction (P18-1)

Challenge:	elaborations are generated using language models that generate background knowledge that helps improve performance . human evaluations show that the quality of the generated ellaborations is high .
Approach:	They propose to finetune smaller language models to generate useful intermediate context . they compare a language model with an answer predictor and generate elaborations . human evaluations show that the quality of the generated ellaborations is high .
Outcome:	The proposed framework outperforms other models on commonsense questions on four commons sense benchmarks.

Copied to clipboard

Wenya Wang, Sinno Jialin Pan

Information Extraction & Text Mining

Challenge:	supervised learning methods have been used for fine-grained opinion analysis but lack of labeled data hinders learning . authors develop a recursive neural network that could reduce domain shift in word level . a recent paper shows that unsupervised methods fail to adapt well across domains .
Approach:	They propose a supervised neural network that reduces domain shift effectively in word level . they treat these relations as invariant "pivot information" across domains to build structural correspondences .
Outcome:	The proposed model reduces domain shift effectively in word level through syntactic relations . it can be used to predict the relation between two adjacent words in the dependency tree .

Are Machines Better at Complex Reasoning? Unveiling Human-Machine Inference Gaps in Entailment Verification (2024.findings-acl)

Copied to clipboard

Soumya Sanyal, Tianyi Xiao, Jiacheng Liu, Wenya Wang, Xiang Ren

From What Is Said to Why It Is Framed: Intent-Aware News Video Understanding (2026.findings-acl)

Challenge:	Existing evidence that humans make numerous inferences to understand discourse and text is not fully understood.
Approach:	They propose to use textual inference datasets with multi-sentence premises to solve the entailment verification problem.
Outcome:	The proposed model outperforms GPT-3.5 and rivals GPL-4 in EV tasks.

Copied to clipboard

Xiangzheng Kong, Minnan Luo, Wenya Wang, Jiaying Wu, Zhi Zeng, Guang Dai

Reinforcing Compositional Retrieval: Retrieving Step-by-Step for Composing Informative Contexts (2025.findings-acl)

Challenge:	Existing verification methods for short-form news videos neglect communicative intent . stylistic presentation and factual manipulation are often intertwined, resulting in shortcut learning .
Approach:	They propose a theory-grounded representation of communicative intent that captures creator stance, audience need activation, and communication strategy.
Outcome:	The proposed framework captures creator stance, audience need activation, and communication strategy.

Copied to clipboard

Quanyu Long, Jianda Chen, Zhengyuan Liu, Nancy F. Chen, Wenya Wang, Sinno Jialin Pan

Information Retrieval

Challenge:	Large Language Models (LLMs) have demonstrated remarkable capabilities across numerous tasks, yet they often rely on external context to handle complex tasks.
Approach:	They propose a tri-encoder sequential retriever that models a Markov Decision Process (MDP) this method decomposes the probability of retrieving a set of elements into a sequence of conditional probabilities and allows each retrieval step to be conditioned on previously selected examples.
Outcome:	The proposed method outperforms baselines and shows that it can handle multiple pieces of evidence or examples.

Exploring Quality and Diversity in Synthetic Data Generation for Argument Mining (2025.emnlp-main)

Copied to clipboard

Jianzhu Bao, Yuqi Huang, Yang Sun, Wenya Wang, Yice Zhang, Bojun Jin, Ruifeng Xu

Static or Dynamic: Towards Query-Adaptive Token Selection for Video Question Answering (2025.emnlp-main)

Challenge:	Argument Mining (AM) is hindered by the scarcity of structure-annotated datasets, which are expensive to create manually.
Approach:	They propose to use quality-oriented synthesis and diversity-oriented approach to generate argumentative texts with diverse topics and argument structures.
Outcome:	The proposed approach significantly improves existing models in full-data and low-resource settings.

Copied to clipboard

Yumeng Shi, Quanyu Long, Wenya Wang

Visual Data in NLP

Challenge:	Existing approaches to compress video inputs ignore the importance of static and dynamic information in long videos, leading to inefficient token usage within limited budgets.
Approach:	They propose a token selection strategy that adaptively adjusts static and dynamic information based on question requirements.
Outcome:	The proposed method achieves performance improvements (up to 5.8%) on multiple video question answering benchmarks.

Training Language Models to Generate Text with Citations via Fine-grained Rewards (2024.acl-long)

Copied to clipboard

Chengyu Huang, Zeqiu Wu, Yushi Hu, Wenya Wang

Quantifying Semantic Emergence in Language Models (2025.acl-long)

Challenge:	Recent Large Language Models (LLMs) are prone to hallucination and their outputs often contain incorrect or unverifiable claims.
Approach:	They propose a training framework using fine-grained rewards to teach LLMs to generate highly supportive and relevant citations while ensuring the correctness of their responses.
Outcome:	The proposed training framework outperforms existing methods on QA datasets and surpasses GPT-3.5-turbo on LLaMA-2-7B.

Copied to clipboard

Hang Chen, Xinyu Yang, Jiaying Zhu, Wenya Wang

Intent Contrastive Learning Based on Multi-view Augmentation for Sequential Recommendation (2025.coling-main)

Challenge:	Existing evaluation methods for large language models (LLMs) focus on coarse-grained text, not providing interpretations for the behavior of finergrained tokens.
Approach:	They propose a quantitative metric to measure large language models’ ability to extract semantics from input tokens.
Outcome:	The proposed metric compares the entropy reduction observed for a sequence of tokens and individual tokens.

Copied to clipboard

Bo Pei, Yingzheng Zhu, Guangjin Wang, Huajuan Duan, Wenya Wu, Fuyong Xu, Yizhao Zhu, Peiyu Liu, Ran Lu

Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing (2025.findings-acl)

Challenge:	Existing work on intent-related models fails to capture long-term dependencies in user behavior and fails to effectively utilize item relevance.
Approach:	They propose a sequential recommendation framework that combine temporal variability with position encoding that has extrapolation properties to encode sequences, thereby expanding the model’s view of user behavior.
Outcome:	The proposed model improves on three real datasets by 0.8% to 14.7% compared to baselines.

Copied to clipboard

Yifan Lu, Jing Li, Yigeng Zhou, Yihui Zhang, Wenya Wang, Xiucheng Li, Meishan Zhang, Fangming Liu, Jun Yu, Min Zhang

Challenge:	Existing knowledge editing methods for large language models (LLMs) suffer from over-editing, where detoxified models reject legitimate queries, compromising overall performance.
Approach:	They propose a toxicity-aware knowledge editing approach that dynamically detects toxic activation patterns during forward propagation and then routes computations through adaptive inter-layer pathways to mitigate toxicity effectively.
Outcome:	The proposed method outperforms existing methods on large language models and enhances the SafeEdit benchmark.

Task-Aware LLM Routing with Multi-Level Task-Profile-Guided Data Synthesis for Cold-Start Scenarios (2026.acl-long)

Copied to clipboard

Hui Liu, Bin Zou, Kecheng Chen, Jie Liu, Wenya Wang, Haoliang Li

Challenge:	Existing routers generalize poorly in cold-start scenarios where in-domain training data is unavailable.
Approach:	They propose a task-type–aware router approach that models query-conditioned cost and performance via latent task-like variables with prior regularization derived from the synthesized task taxonomy.
Outcome:	The proposed framework improves performance and cost under cold-start and in-domain settings and enables efficient routing.

Personalizing LLMs with Binary Feedback: A Preference-Calibrated Optimization Framework (2026.acl-long)

Copied to clipboard

Xilai Ma, Liye Zhao, Weijun Yao, Haibing Di, Wenya Wang, Jing Li

Challenge:	Existing methods focus on isolated user histories, neglecting the essential role of inter-user differences.
Approach:	They propose a framework that personalizes Large Language Models via preference-calibrated binary signals.
Outcome:	The proposed framework outperforms baselines in a variety of personalization tasks and backbone LLMs.

TELLER: A Trustworthy Framework for Explainable, Generalizable and Controllable Fake News Detection (2024.findings-acl)

Copied to clipboard

Hui Liu, Wenya Wang, Haoru Li, Haoliang Li

MLLM-Bench: Evaluating Multimodal LLMs with Per-sample Criteria (2025.naacl-long)

Challenge:	Existing methods for detecting fake news are limited due to non-transparent reasoning processes and inherent risks of integration with large language models.
Approach:	They propose a framework for trustworthy fake news detection that prioritizes explainability, generalizability and controllability of models.
Outcome:	The proposed framework prioritizes explainability, generalizability and controllability of models.

Copied to clipboard

Wentao Ge, Shunian Chen, Hardy Chen, Nuo Chen, Junying Chen, Zhihong Chen, Wenya Xie, Shuo Yan, ChenghaoZhu ChenghaoZhu, Ziyue Lin, Dingjie Song, Xidong Wang, Anningzhe Gao, Zhang Zhiyi, Jianquan Li, Xiang Wan, Benyou Wang

HyperPELT: Unified Parameter-Efficient Language Model Tuning for Both Language and Vision-and-Language Tasks (2023.findings-acl)

Challenge:	Existing evaluation methodologies for multimodal large language models are limited in evaluating objective queries without considering real-world user experiences.
Approach:	They propose to evaluate multimodal large language models with per-sample criteria using potent MLLM as the judge.
Outcome:	The proposed evaluation paradigm shows that it can be used to evaluate multimodal large language models with per-sample criteria.

Copied to clipboard

Zhengkun Zhang, Wenya Guo, Xiaojun Meng, Yasheng Wang, Yadao Wang, Xin Jiang, Qun Liu, Zhenglu Yang

Multi-Modality Expansion and Retention for LLMs through Parameter Merging and Decoupling (2025.acl-long)

Challenge:	Pretraining and fine-tuning are the dominant paradigms in natural language processing.
Approach:	They propose a parameter-efficient multitask learning framework that takes trainable hyper-embeddings and visual modality as input and outputs weights for different modules in a pretrained language model.
Outcome:	The proposed framework adds fewer trainable parameters in multi-task learning while achieving superior performances and transfer ability compared to state-of-the-art methods.

Copied to clipboard

Junlin Li, Guodong Du, Jing Li, Sim Kuan Goh, Wenya Wang, Yequan Wang, Fangming Liu, Ho-Kin Tang, Saleh Alharbi, Daojing He, Min Zhang

Coordinating Search-Informed Reasoning and Reasoning-Guided Search in Claim Verification (2026.acl-long)

Challenge:	Large Language Models (LLMs) are a cornerstone in artificial intelligence due to their exceptional performance.
Approach:	They propose a training-free approach that integrates existing MLLMs for effective multimodal expansion while retaining their original performance.
Outcome:	The proposed approach can expand LLMs' multimodal capabilities while retaining original performance.

Copied to clipboard

Qisheng Hu, Quanyu Long, Wenya Wang

Towards Multi-Modal Sarcasm Detection via Hierarchical Congruity Modeling with Knowledge Enhancement (2022.emnlp-main)

Challenge:	Existing methods for multi-hop claim verification require multi-step reasoning to construct verification chains while iterating for information to uncover hidden bridging facts.
Approach:	They propose a hierarchical agent reasoning and information search model that integrates reasoning and search-informed reasoning.
Outcome:	Experimental results show that HARIS improves multi-hop claim verification accuracy and interpretability.

Copied to clipboard

Hui Liu, Wenya Wang, Haoliang Li

Domain Confused Contrastive Learning for Unsupervised Domain Adaptation (2022.naacl-main)

Challenge:	Sarcasm is a linguistic phenomenon indicating a discrepancy between literal meanings and implied intentions.
Approach:	They propose a hierarchical framework for sarcasm detection by exploring atomic-level congruity and composition-level convergence.
Outcome:	The proposed model outperforms existing methods on a public sarcasm detection dataset based on Twitter .

Copied to clipboard

Quanyu Long, Tianze Luo, Wenya Wang, Sinno Pan

Jianzhu Bao, Haozhen Zhang, Kuicai Dong, Bozhi Wu, Sarthak Ketanbhai Modi, Zi Pong Lim, Yon Shin Teo, Wenya Wang

Challenge:	Existing studies on domain-shifting adaptations have focused on domain .
Approach:	They propose a self-supervised approach to unsupervised domain adduction using domain puzzles to bridge the source and target domains and retain discriminative representations after adaptation.
Outcome:	The proposed approach outperforms baselines and further ablation studies show that it is more stable and effective when performing other data augmentations.

Copied to clipboard

Language Models over Large-Scale Knowledge Base: on Capacity, Flexibility and Reasoning for New Facts (2025.coling-main)

Challenge:	Chart understanding is a critical capability for vision-language models, serving as a cornerstone for automated data analysis, document understanding, and scientific research.
Approach:	They propose a chart-efficient training framework to enhance counterfactual sensitivity by code modification and a similarity-based data selection strategy.
Outcome:	The proposed framework achieves superior or comparable performance to strong chart-specific VLMs while using significantly less training data.

Copied to clipboard

Qiyuan He, Yizhong Wang, Jianfei Yu, Wenya Wang

From Competition to Synergy: Unlocking Reinforcement Learning for Subject-Driven Image Generation (2026.acl-long)

Challenge:	Existing studies on LMs lack systematic studies on their structured reasoning capabilities over the infused knowledge.
Approach:	They investigate how LMs of different sizes can store world knowledge of different frequencies in a large-scale KB after training on the abundant world knowledge triplets.
Outcome:	The proposed models can store and respond to natural language queries with flexibility and reasoning abilities, but they need to be enhanced to fully realize their potential.

Copied to clipboard

Ziwei Huang, Ying Shu, null Fanghao, Quanyu Long, Wenya Wang, Qiushi Guo, Tiezheng Ge, Leilei Gan

Unraveling the Mechanics of Learning-Based Demonstration Selection for In-Context Learning (2025.acl-long)

Challenge:	a naive application of GRPO leads to conflicting gradient signals and a misalignment with the temporal dynamics of the diffusion process.
Approach:	They propose a framework that uses synergy-aware reward shaping to penalize conflicted reward signals and amplify synergies to provide a sharper and decisive gradient.
Outcome:	The proposed framework outperforms naive GRPO and Time-Aware Dynamic Weighting (TDW) on DreamBench, and achieves a state-of-the-art balance between ID preservation and prompt adherence.

Copied to clipboard

Hui Liu, Wenya Wang, Hao Sun, Chris Xing Tian, Chenqi Kong, Xin Dong, Haoliang Li

Programming over Thinking: Efficient and Robust Multi-Constraint Planning (2026.acl-long)

Challenge:	Recent learning-based demonstration selection methods have proven beneficial to in-context learning (ICL) by choosing more useful exemplars.
Approach:	They propose two methods to capture task-agnostic similarities between input and output of LLMs.
Outcome:	The proposed methods integrate task-agnostic similarities of different levels between input and output of exemplars and test cases to eliminate costly data collection.

Copied to clipboard

Derrick Goh Xin Deik, Quanyu Long, Zhengyuan Liu, Nancy F. Chen, Wenya Wang

Deep Inductive Logic Reasoning for Multi-Hop Reading Comprehension (2022.acl-long)

Challenge:	Existing large language model approaches lack flexibility in multi-constraint planning . SCOPE achieves state-of-the-art performance while lowering cost and latency .
Approach:	They propose a framework that disentangles query-specific problem reasoning from generic code execution.
Outcome:	The Scalable Code Planning Engine achieves state-of-the-art performance while lowering cost and latency.

Copied to clipboard

Wenya Wang, Sinno Pan

Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination (2026.acl-long)

Challenge:	a new method for multi-hop reading comprehension uses deep learning to reason across documents . a previous study showed that deep learning methods only implicitly encode query-related information into embeddings .
Approach:	They propose a deep-learning based inductive logic reasoning method that extracts query-related information and conducts logic reasoning among filtered information.
Outcome:	The proposed model is evaluated on two reading comprehension datasets . it uses attentive memories with novel differentiable logic operators .

Copied to clipboard

Yangneng Chen, Junlin Li, Weijun Yao, Xilai Ma, Guodong DU, Wenya Wang, Jing Li

Policy-Guided Stepwise Action Planning for Controllable LLM Reasoning (2026.findings-acl)

Challenge:	Large Vision-Language Models (LVLMs) are capable of processing visual inputs, but are susceptible to hallucinations.
Approach:	They propose a method to localize and localize specific visual tokens, which are defined as Inert Tokens, across layers, revealing a rigid semantic collapse.
Outcome:	The proposed approach reduces the likelihood of LVLMs being hijacked by visual inputs while maintaining general capabilities.

Copied to clipboard

Jianpeng Zhou, Qisheng Hu, Jiahai Wang, Wenya Wang

Vera: A General-Purpose Plausibility Estimation Model for Commonsense Statements (2023.emnlp-main)

Challenge:	Existing approaches to steering large language model reasoning via high-level reasoning actions fail to outperform standard generation because planners tend to degenerate into repetitive loops or fixed patterns.
Approach:	They propose a planner-executor framework that learns to select reasoning actions dynamically while keeping the executor LLM fully frozen.
Outcome:	The proposed framework outperforms existing paradigms by preserving the executor LLM frozen . PG-HAP improves accuracy over strong baselines while producing less redundant, more adaptive trajectories.

Copied to clipboard

Jiacheng Liu, Wenya Wang, Dianzhuo Wang, Noah Smith, Yejin Choi, Hannaneh Hajishirzi

Commonsense Reasoning

Challenge:	Despite the advances of language models, they still produce text that contains trivial commonsense errors.
Approach:	They propose a general-purpose commonsense statement verification model that learns to estimate the plausibility of declarative statements based on commonsensical knowledge.
Outcome:	The proposed model outperforms existing models that can be repurposed for commonsense verification, even including GPT-3.5/ChatGPT/GPT-4.

A Span-based Multimodal Variational Autoencoder for Semi-supervised Multimodal Named Entity Recognition (2022.emnlp-main)

Copied to clipboard

Baohang Zhou, Ying Zhang, Kehui Song, Wenya Guo, Guoqing Zhao, Hongbin Wang, Xiaojie Yuan

Interpretable Multimodal Misinformation Detection with Logic Reasoning (2023.findings-acl)

Challenge:	Existing methods for named entity recognition on social media are not efficient for semi-supervised MNER because of the mismatch between the posted text and image.
Approach:	They propose a novel method to fuse the text and image features for multimodal named entity recognition under semi-supervised setting by exploiting modal-specific VAEs.
Outcome:	The proposed method outperforms baselines under supervised setting and improves performance with less labeled data than existing semi-supervised methods.

Copied to clipboard

Hui Liu, Wenya Wang, Haoliang Li

STARE at the Structure: Steering ICL Exemplar Selection with Structural Alignment (2025.emnlp-main)

Challenge:	Existing methods for misinformation detection lack interpretability due to the black-box nature of the neural network.
Approach:	They propose a logic-based neural model which integrates interpretable logic clauses to express the reasoning process of the target task.
Outcome:	The proposed model can be generalizable across multiple misinformation sources and is based on three public datasets.

Copied to clipboard

Jiaqian Li, Qisheng Hu, Jing Li, Wenya Wang