Papers by Fang Wei

Context-DPO: Aligning Language Models for Context-Faithfulness (2025.findings-acl)

Challenge:	Large language models (LLMs) are evolving from text generation into integration within agentic workflows . tools such as APIs, databases, and software tools are expanding rapidly .
Approach:	They propose a lightweight framework that models retrieval as iterative query planning . instead of single-shot matching, ToolQP decomposes instructions into sub-tasks .
Outcome:	The proposed framework achieves state-of-the-art performance and robustness across retrievers.

Copied to clipboard

Baolong Bi, Shaohan Huang, Yiwei Wang, Tianchi Yang, Zihan Zhang, Haizhen Huang, Lingrui Mei, Junfeng Fang, Zehao Li, Furu Wei, Weiwei Deng, Feng Sun, Qi Zhang, Shenghua Liu

On the Role of Entity and Event Level Conceptualization in Generalizable Reasoning: A Survey of Tasks, Methods, Applications, and Future Directions (2025.findings-emnlp)

Challenge:	Context-DPO is the first alignment method specifically designed to enhance contextfaithfulness for large language models.
Approach:	They propose a benchmark that simulates Retrieval-Augmented Generation scenarios with knowledge conflicts to evaluate context-faithfulness.
Outcome:	The proposed method improves LLMs' context-faithfulness by 35% to 280% over open-source models.

Copied to clipboard

Weiqi Wang, Tianqing Fang, Haochen Shi, Baixuan Xu, Wenxuan Ding, Liyu Zhang, Wei Fan, Jiaxin Bai, Haoran Li, Xin Liu, Yangqiu Song

Reasoning

Challenge:	Conceptualization is a fundamental element of human cognition and plays a pivotal role in generalizable reasoning.
Approach:	They propose to categorize different types of conceptualizations into four levels based on the types of instances being conceptualized.
Outcome:	The proposed categorization of different types of conceptualizations into four levels based on the types of instances being conceptualized .

Modeling Adversarial Attack on Pre-trained Language Models as Sequential Decision Making (2023.findings-acl)

Copied to clipboard

Xuanjie Fang, Sijie Cheng, Yang Liu, Wei Wang

Robustness in NLP

Challenge:	Pre-trained language models (PLMs) have shown strong potential in various downstream tasks.
Approach:	They propose to model adversarial attack task as a sequential decision-making problem where the whole attack process is sequential with two decision- making problems, i.e., word finder and word substitution.
Outcome:	The proposed approach achieves the highest attack success rate with a comparable modification rate and semantic similarity to attack fine-tuned BERT.

Infinity-Parser: Layout-Aware Reinforcement Learning with High-quality Document Parsing Dataset (2026.findings-acl)

Copied to clipboard

Baode Wang, Biao Wu, Weizhen Li, Meng Fang, Zuming Huang, Jun Huang, Yanjie Liang, Haozhe Wang, Ling Chen, Wei Chu, Yuan Qi

Multimodality

Challenge:	Existing supervised fine-tuning methods struggle to generalize across document types, leading to poor performance.
Approach:	They propose layoutRL, a reinforcement learning framework that optimizes layout understanding through composite rewards integrating normalized edit distance, paragraph count accuracy, and reading order preservation.
Outcome:	The proposed model outperforms specialized document parsing systems and general-purpose vision-language models on a broad range of document types, languages, and structural complexities.

Watermarking Large Language Models: An Unbiased and Low-risk Method (2025.acl-long)

Copied to clipboard

Minjia Mao, Dongjun Wei, Zeyu Chen, Xiao Fang, Michael Chau

Language Models

Challenge:	Recent advances in large language models (LLMs) have highlighted the risk of misusing them, raising the need for accurate detection of LLM-generated content.
Approach:	They propose a method to inject imperceptible identifiers into large language models (LLMs) this method is unbiased and preserves the original token distribution in expectation .
Outcome:	The proposed method preserves the original token distribution in expectation and has lower risk of producing unsatisfactory outputs in low-entropy scenarios compared to existing unbiased watermarks.

Does Reasoning Introduce Bias? A Study of Social Bias Evaluation and Mitigation in LLM Reasoning (2025.findings-emnlp)

Copied to clipboard

Xuyang Wu, Jinming Nian, Ting-Ruen Wei, Zhiqiang Tao, Hsin-Tai Wu, Yi Fang

Structural Bias for Aspect Sentiment Triplet Extraction (2022.coling-1)

Challenge:	Recent advances in large language models have enabled automatic generation of chain-of-thought reasoning . however, when reasoning steps reflect social stereotypes, they can reinforce harmful associations and lead to misleading conclusions.
Approach:	They propose a method that detects how model predictions change across incremental reasoning steps.
Outcome:	The proposed method outperforms a stereotype-free baseline and improves accuracy.

Copied to clipboard

Chen Zhang, Lei Ren, Fang Ma, Jingang Wang, Wei Wu, Dawei Song

Sentiment Analysis

Challenge:	Existing structural bias adapters for aspect sentiment triplet extraction are under-confident . a large-scale dataset for ASTE shows the adapter is effective and efficient to a larger scale.
Approach:	They propose to use a structural adapter to integrate structural bias into pretrained language models . they propose to add a relative position structure in place of the syntactic dependency structure .
Outcome:	The proposed adapter achieves state-of-the-art performance over strong baselines, but with a light parameter demand and low latency.

BotChat: Evaluating LLMs’ Capabilities of Having Multi-Turn Dialogues (2024.findings-naacl)

Copied to clipboard

Haodong Duan, Jueqi Wei, Chonghua Wang, Hongwei Liu, Yixiao Fang, Songyang Zhang, Dahua Lin, Kai Chen

Dialogue Systems & Conversational Agents

Challenge:	Modern Large Language Models (LLMs) facilitate high-quality, multi-turn dialogues with humans, but human-based evaluation of such a capability requires substantial manual effort.
Approach:	They propose to evaluate LLMs' ability to emulate human-like, multi-turn conversations using an LLM-centric approach.
Outcome:	The proposed model emulates human-like, multi-turn conversations using an LLM-centric approach.

Where to Attack: A Dynamic Locator Model for Backdoor Attack in Text Classifications (2022.coling-1)

Copied to clipboard

Heng-yang Lu, Chenyou Fan, Jun Yang, Cong Hu, Wei Fang, Xiao-jun Wu

Robustness in NLP

Challenge:	BackDoor Attack (BDA) study aims to train a poisoned model with clean data and some trigger-embedded instances to perform normally on normal inputs.
Approach:	They propose to train a poisoned model with clean and poisonest inputs . they propose to use triggers to predict those poisonets as target labels .
Outcome:	The proposed model can predict P2P dynamically without human intervention.

Neural Multi-Task Learning for Stance Prediction (D19-66)

Copied to clipboard

Wei Fang, Moin Nadeem, Mitra Mohtarami, James Glass

Forget the Token and Pixel: Rethinking Gradient Ascent for Concept Unlearning in Multimodal Generative Models (2025.findings-acl)

Challenge:	Existing models for fact checking are limited in size due to limited data available . stance detection is a key component of fact checking for journalists and news agencies .
Approach:	They propose to use textual information from existing datasets to improve stance prediction.
Outcome:	The proposed model outperforms state-of-the-art systems on a public benchmark dataset by 6.0 and 14.4 points in weighting.

Copied to clipboard

Jiaqi Li, Chuanyi Zhang, Miaozeng Du, Hui Zhang, Yongrui Chen, Qianshan Wei, Junfeng Fang, Ruipeng Wang, Sheng Bi, Guilin Qi

Beyond Quantity: Trajectory Diversity Scaling for Code Agents (2026.findings-acl)

Challenge:	Gradient Ascent (GA) has emerged as a promising approach for concept unlearning in Multimodal Generative Models (MGMs).
Approach:	They propose a novel approach that selectively applies GA to targeted Conceptual Knowledge while preserving Natural Knowledge through Gradient Descent (GD).
Outcome:	The proposed approach removes Conceptual Knowledge and inadvertently diminishes Natural Knowledge, resulting in utility degradation.

Copied to clipboard

Guhong Chen, Chenghao Sun, Cheng Fu, Qiyao Wang, Zhihong Huang, ChaoPeng Wei, Guangxu Chen, Feiteng Fang, Ahmadreza Argha, Bing Zhao, Xander Xu, Qi Han, Hamid Alinejad-Rokny, Qiang Qu, Binhua Li, Shiwen Ni, Min Yang, HU Wei, Yongbin Li

Green & Sustainable NLP

Challenge:	Code large language models (LLMs) are becoming tool-interactive agents . quantity-centric scaling exhibits an early bottleneck that underutilizes trajectory data . et al.: a new approach to scale trajectory diversity improves tool-use generalization .
Approach:	They propose a Trajectory Diversity Scaling-based data synthesis framework for code agents that scales performance through diversity rather than raw volume.
Outcome:	Experiments on general tool-use benchmarks and code agent tasks show that TDScaling improves tool-user generalization and inherent coding proficiency.

MPL: Multiple Programming Languages with Large Language Models for Information Extraction (2025.findings-acl)

Copied to clipboard

Bo Li, Gexiang Fang, Wei Ye, Zhenghua Xu, Jinglei Zhang, Hao Cheng, Shikun Zhang

Pretrained Language Models for Dialogue Generation with Multiple Input Sources (2020.findings-emnlp)

Challenge:	Existing research focuses on Python for code-style simulation, overlooking the potential of other widely-used PLs during the supervised fine-tuning phase.
Approach:	They propose a framework that incorporates programming languages into IE tasks . they introduce function-prompt with virtual running to simulate code-style inputs .
Outcome:	The proposed framework exploits the potential of different programming languages during the supervised fine-tuning phase.

Copied to clipboard

Yu Cao, Wei Bi, Meng Fang, Dacheng Tao

Dialogue Systems & Conversational Agents

Challenge:	Large-scale pretrained language models have achieved outstanding performance on natural language understanding tasks.
Approach:	They propose to fuse attention information from multiple input sources to achieve better relevance with dialogue history than simple fusion baselines.
Outcome:	The proposed models deliver higher relevance with dialogue history than baselines.

FAKTA: An Automatic End-to-End Fact Checking System (N19-4)

Copied to clipboard

Moin Nadeem, Wei Fang, Brian Xu, Mitra Mohtarami, James Glass

Ethical NLP

Challenge:	Existing studies have investigated individual components of fact checking process but none offer such a capability.
Approach:	They propose a framework that integrates various components of a fact-checking process.
Outcome:	The proposed framework integrates various components of a fact-checking process to predict the factuality of claims and provide evidence at the document and sentence level to explain its predictions.

Expand, Rerank, and Retrieve: Query Reranking for Open-Domain Question Answering (2023.findings-acl)

Copied to clipboard

Yung-Sung Chuang, Wei Fang, Shang-Wen Li, Wen-tau Yih, James Glass

PLAY2PROMPT: Zero-shot Tool Instruction Optimization for LLM Agents via Tool Play (2025.findings-acl)

Challenge:	Empirically, EAR improves top-5/20 accuracy by 3-8 and 5-10 points . dense retrievers are limited by their inability to perform semantic matching for relevant passages that have low lexical overlap with the query.
Approach:	They propose a query expansion and reranking approach for improving passage retrieval with the application to open-domain question answering.
Outcome:	Empirically, EAR improves top-5/20 accuracy by 3-8 and 5-10 points when compared to a vanilla query expansion model and a dense retrieval model.

Copied to clipboard

Wei Fang, Yang Zhang, Kaizhi Qian, James R. Glass, Yada Zhu

Language Models

Challenge:	Existing solutions for large language models rely on manual rewriting or labeled data for validation . Existing approaches rely only on comprehensive tool documentation and in-context demonstrations .
Approach:	They propose a framework that "plays" with each tool to explore its input-output behaviors.
Outcome:	Experiments show that PLAY2PROMPT improves zero-shot tool performance across open and closed models.

HCL-TAT: A Hybrid Contrastive Learning Method for Few-shot Event Detection with Task-Adaptive Threshold (2022.findings-emnlp)

Copied to clipboard

Ruihan Zhang, Wei Wei, Xian-Ling Mao, Rui Fang, Dangyang Chen

Retrieve-and-Sample: Document-level Event Argument Extraction via Hybrid Retrieval Augmentation (2023.acl-long)

Challenge:	Existing methods for event detection have failed to address the problem of constantly emerging event types with limited data.
Approach:	They propose a novel method for event detection with a task-adaptive threshold . they propose to learn discriminative representations with 'two-view contrastive loss'
Outcome:	The proposed method achieves better results than the state-of-the-art methods on a benchmark dataset.

Copied to clipboard

Yubing Ren, Yanan Cao, Ping Guo, Fang Fang, Wei Ma, Zheng Lin

EduMARS: Can Vision-Language Models Grade Like Teachers? Benchmarking Multimodal, Rubric-Based Assessment on Chinese K-12 Answers (2026.findings-acl)

Challenge:	Recent studies show the effectiveness of retrieval augmentation in many generative NLP tasks.
Approach:	They investigate retrieval settings from the input and label distribution views . they further augment document-level EAE with pseudo demonstrations sampled from event semantic regions .
Outcome:	The proposed methods can augment document-level EAE with pseudo demonstrations . the methods can be used in generative NLP tasks such as dialogue response generation .

Copied to clipboard

Xuan Zhao, Jiashun Chen, Wanting xu, Huiyuan Yan, Chaowei Fang, Xing Wei

Multimodality

Challenge:	Existing benchmarks for automated grading of student work fail to evaluate real student responses . existing models fail to assess real student work, especially on cognitively demanding tasks .
Approach:	They propose a multimodal benchmark for rubric-aligned evaluation of real Chinese K-12 student answers.
Outcome:	The proposed model improves performance and interpretability of existing models on EduMARS . existing models fail to perform on real-world, cognitively demanding tasks, authors say .

NL2Lean: Translating Natural Language into Lean 4 through Multi-Aspect Reinforcement Learning (2025.emnlp-main)

Copied to clipboard

Yue Fang, Shaohan Huang, Xin Yu, Haizhen Huang, Zihan Zhang, Weiwei Deng, Furu Wei, Feng Sun, Qi Zhang, Zhi Jin

Machine Translation

Challenge:	Existing formal proof assistants rely on instruction tuning and lack fine-grained structural and semantic alignment.
Approach:	They propose a reinforcement learning framework that enables LLMs to translate natural language into formal language such as Lean 4 . they use a model with basic translation ability to refine the model's reinforcement learning .
Outcome:	The proposed method outperforms baseline models on NL-to-Lean 4 tasks.

Joint Inference of Retrieval and Generation for Passage Re-ranking (2024.findings-eacl)

Copied to clipboard

Wei Fang, Yung-Sung Chuang, James Glass

XPrompt: Exploring the Extreme of Prompt Tuning (2022.emnlp-main)

Challenge:	Existing methods for re-ranking documents are sparse and do not require training.
Approach:	They propose a method that optimizes mutual information between query and passage distributions by integrating cross-encoders and generative models in the re-ranking process.
Outcome:	The proposed method outperforms conventional re-rankers and language model scorers in open-domain QA retrieval settings and diverse retrieval benchmarks under zero-shot settings.

Copied to clipboard

Fang Ma, Chen Zhang, Lei Ren, Jingang Wang, Qifan Wang, Wei Wu, Xiaojun Quan, Dawei Song

Not All Voices Are Rewarded Equally: Probing and Repairing Reward Models across Human Diversity (2025.findings-emnlp)

Challenge:	Prompt tuning learns soft prompts to condition pre-trained Language Models for performing downstream tasks in a parameter-efficient manner.
Approach:	They propose a Prompt tuning model with an eXtremely small scale that learns soft prompts to condition the frozen Pre-trained Language Models for performing downstream tasks in a parameter-efficient manner.
Outcome:	The proposed model outperforms the vanilla Prompt-Tuning and can significantly improve across tasks and model scales.

Copied to clipboard

Zihao Li, Feihao Fang, Xitong Zhang, Jiaru Zou, Zhining Liu, Wei Xiong, Ziwei Wu, Baoyu Jing, Jingrui He

Multimodality

Challenge:	Using real-world datasets, we conduct the most comprehensive study to date, auditing various state-of-the-art reward models across nine sensitive attributes, including age, gender, ethnicity, etc.
Approach:	They propose a method to mitigate group disparities in reward modeling by using real-world data.
Outcome:	The proposed method is based on a population-based dataset with nine demographic attributes, including gender, ethnicity, age, gender, and ethnicity.

AbsInstruct: Eliciting Abstraction Ability from LLMs through Explanation Tuning with Plausibility Estimation (2024.acl-long)

Copied to clipboard

Zhaowei Wang, Wei Fan, Qing Zong, Hongming Zhang, Sehyun Choi, Tianqing Fang, Xin Liu, Yangqiu Song, Ginny Wong, Simon See

Explainability & Interpretability in NLP

Challenge:	Existing work shows that LLMs are deficient in abstract ability, and how to improve it remains unexplored.
Approach:	They propose a framework AbsInstruct to enhance LLMs’ abstract ability through instruction tuning.
Outcome:	The proposed framework can enhance LLMs’ abstraction ability with strong generalization performance while maintaining their general instruction-following abilities.

DEIE: Benchmarking Document-level Event Information Extraction with a Large-scale Chinese News Dataset (2024.lrec-main)

Copied to clipboard

Yubing Ren, Yanan Cao, Hao Li, Yingjie Li, Zixuan ZM Ma, Fang Fang, Ping Guo, Wei Ma

Information Extraction & Text Mining

Challenge:	Existing event-based datasets mainly target sentence-level tasks . current models struggle with "document" annotation, a key feature of the current model .
Approach:	They propose a large-scale document-level event information extraction dataset with over 56,000+ events and 242,000+ arguments.
Outcome:	The proposed dataset has over 56,000+ events and 242,000+ arguments.

A Model-agnostic Data Manipulation Method for Persona-based Dialogue Generation (2022.acl-long)

Copied to clipboard

Yu Cao, Wei Bi, Meng Fang, Shuming Shi, Dacheng Tao

Dialogue Systems & Conversational Agents

Challenge:	Existing models for introducing explicit personas are expensive due to their expensive collection costs.
Approach:	They propose a data manipulation method which is model-agnostic to be packed with any persona-based dialogue generation model to improve their performance.
Outcome:	The proposed method is model-agnostic to be packed with any persona-based dialogue generation model to improve their performance.

Who Can Withstand Chat-Audio Attacks? An Evaluation Benchmark for Large Audio-Language Models (2025.findings-acl)

Copied to clipboard

Wanqi Yang, Yanda Li, Meng Fang, Yunchao Wei, Ling Chen

Robustness in NLP

Challenge:	Existing research focused on model-specific adversarial methods, but real-world applications demand a more generalizable approach to audio adversarials.
Approach:	They propose a Chat-Audio Attacks benchmark to evaluate LALMs' robustness . they propose standard evaluation, GPT-4o-based evaluation and human evaluation .
Outcome:	The proposed benchmark aims to explore the robustness of six state-of-the-art LALMs with voice interaction capabilities.

Joint Pre-Encoding Representation and Structure Embedding for Efficient and Low-Resource Knowledge Graph Completion (2024.emnlp-main)

Copied to clipboard

Chenyu Qiu, Pengjiang Qian, Chuang Wang, Jian Yao, Li Liu, Fang Wei, Eddie Eddie

CLIO: Role-interactive Multi-event Head Attention Network for Document-level Event Extraction (2022.coling-1)

Challenge:	Existing knowledge graph completion models require longer training and inference times as well as increased memory usage.
Approach:	They propose to encode textual descriptions into semantic representations before training and integrate structural embedding with pre-encoded semantic description to improve model's prediction performance on 1-N relations.
Outcome:	The proposed model increases inference speed by 30x and reduces training memory by approximately 60% on the WN18RR and UMLS datasets.

Copied to clipboard

Yubing Ren, Yanan Cao, Fang Fang, Ping Guo, Zheng Lin, Wei Ma, Yi Liu

Information Extraction & Text Mining

Challenge:	Existing methods for document-level event extraction struggle due to two intrinsic challenges: nested arguments and multiple events.
Approach:	They propose a role-interactive multi-event head attention network to solve two challenges . they map different events to multiple subspaces and then determine whether the current event exists .
Outcome:	The proposed model improves on two widely used DEE datasets on the Internet.

Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning (2026.acl-long)

Copied to clipboard

Bo Li, Mingda Wang, Gexiang Fang, Shikun Zhang, Wei Ye

Train in Vain: Functionality-Preserving Poisoning to Prevent Unauthorized Use of Code Datasets (2026.findings-acl)

Challenge:	Existing models that ground retrieval on external evidence are limited in their ability to implement retrieval-augmented generation.
Approach:	They propose a retrieval-augmented generation model that embeds retrieval control directly into generation.
Outcome:	The proposed model surpasses strong RAG baselines and uses substantially fewer parameters.

Copied to clipboard

Yuan Xiao, Jiaming Wang, Yuchen Chen, Wei Song, Jun Sun, Shiqing Ma, Yanzhou Mu, Juan Zhai, Chunrong Fang, Jin Song Dong, Zhenyu Chen

Programming Languages in NLP

Challenge:	Existing methods for dataset poisoning require full-dataset poison, which breaks code compilability.
Approach:	They propose a functionality-preserving poisoning approach that injects short, compilable weak-use fragments into executed code paths.
Outcome:	The proposed method contaminates 10% of the dataset while maintaining 100% compilability and functional correctness.

Debiasing LLMs by Masking Unfairness-Driving Attention Heads (2026.findings-acl)

Copied to clipboard

Tingxu Han, Wei Song, Ziqi Ding, Ziming Li, Chunrong Fang, Yuekang Li, Dongfang Liu, Zhenyu Chen, Zhenting Wang