Papers by Liang Pang

Media Source Matters More Than Content: Unveiling Political Bias in LLM-Generated Citations (2025.emnlp-main)

Challenge:	Existing defenses rely on impractical assumptions about trigger settings to mitigate backdoor attacks . a recent study found that small amounts of training data can systematically induce harmful behaviors in large language models.
Approach:	They propose a backdoor defense framework that requires no prior knowledge of trigger settings . they use a two-stage process to aggregate backdoor representations and fine-tune recovery .
Outcome:	The proposed defense reduces the average Attack Success Rate to 4.41% across multiple benchmarks . the proposed framework generalizes across different types of backdoors, confirming its robustness in practical deployment scenarios.

Copied to clipboard

Sunhao Dai, Zhanshuo Cao, Wenjie Wang, Liang Pang, Jun Xu, See-Kiong Ng, Tat-Seng Chua

MLaKE: Multilingual Knowledge Editing Benchmark for Large Language Models (2025.coling-main)

Challenge:	generative search engines rely on in-line citations as the key gateway to original webpages . a recent study shows that LLMs tend to cite left-leaning sources at higher rates compared to traditional retrieval systems .
Approach:	They construct a dataset of news articles labeled with left- or right-leaning stances . they find that LLMs tend to cite left-leansing sources at higher rates than traditional retrieval systems .
Outcome:	The proposed dataset shows that LLMs tend to cite left-leaning sources at higher rates than traditional retrieval systems.

Copied to clipboard

Zihao Wei, Jingcheng Deng, Liang Pang, Hanxing Ding, Huawei Shen, Xueqi Cheng

Beyond Memorization: The Challenge of Random Memory Access in Language Models (2024.acl-long)

Challenge:	Existing studies on knowledge editing focus on monolingual scenarios, neglecting the complexities presented by multilingual contexts and multi-hop reasoning.
Approach:	They propose a benchmark to evaluate the adaptability of multilingual knowledge editing methods.
Outcome:	The proposed benchmark evaluates the adaptability of multilingual knowledge editing methods across five languages.

Copied to clipboard

Tongyao Zhu, Qian Liu, Liang Pang, Zhengbao Jiang, Min-Yen Kan, Min Lin

Cocktail: A Comprehensive Information Retrieval Benchmark with LLM-Generated Documents Integration (2024.findings-acl)

Challenge:	Recent advances in Language Models (LMs) have shown their effectiveness in knowledge-intensive tasks.
Approach:	They investigate whether a generative language model is able to access its memory sequentially or randomly.
Outcome:	The proposed LMs are able to access memory sequentially or randomly.

Copied to clipboard

Sunhao Dai, Weihao Liu, Yuqi Zhou, Liang Pang, Rongju Ruan, Gang Wang, Zhenhua Dong, Jun Xu, Ji-Rong Wen

Think Before You Speak: Cultivating Communication Skills of Large Language Models via Inner Monologue (2024.findings-naacl)

Challenge:	Large Language Models (LLMs) have led to an influx of AI-generated content on the internet, transforming corpus of Information Retrieval (IR) systems from human-written to a coexistence with LLM-generated contents.
Approach:	They propose a benchmark named Cocktail that compares IR models with LLMs to find relevant documents and passages from a corpus.
Outcome:	The proposed benchmark aims to evaluate IR models in the mixed-sourced data landscape of the LLM era.

Copied to clipboard

Junkai Zhou, Liang Pang, Huawei Shen, Xueqi Cheng

Wasserstein Distance Regularized Sequence Representation for Text Matching in Asymmetrical Domains (2020.emnlp-main)

Challenge:	emergence of large language models (LLMs) improves capabilities of dialogue systems . but they lack communication skills, which make them more like information seeking tools .
Approach:	They propose to empower LLMs with communication skills through inner monologues . they use a benchmark to evaluate the dialogue generation ability of the model .
Outcome:	The proposed model outperforms the baselines in the evaluation of communication skills.

Copied to clipboard

Weijie Yu, Chen Xu, Jun Xu, Liang Pang, Xiaopeng Gao, Xiaozhao Wang, Ji-Rong Wen

Representation Learning

Challenge:	Asymmetrical text matching is a fundamental problem in information retrieval and natural language processing.
Approach:	They propose a method that regularizes features vectors projected from different domains . WD-Match can be used to improve different text matching methods .
Outcome:	The proposed method outperforms existing methods and benchmarks on four datasets.

BERM: Training the Balanced and Extractable Representation for Matching to Improve Generalization Ability of Dense Retrieval (2023.acl-long)

Copied to clipboard

Shicheng Xu, Liang Pang, Huawei Shen, Xueqi Cheng

ToolCoder: A Systematic Code-Empowered Tool Learning Framework for Large Language Models (2025.acl-long)

Challenge:	Dense retrieval has shown promise in the first-stage retrieval process when trained on in-domain labeled datasets.
Approach:	They propose a method to capture matching signal to improve generalization of dense retrieval by capturing matching signal between two texts.
Outcome:	The proposed method can be combined with different training methods to improve generalization ability without additional inference overhead and target domain data.

Copied to clipboard

Hanxing Ding, Shuchang Tao, Liang Pang, Zihao Wei, Jinyang Gao, Bolin Ding, Huawei Shen, Xueqi Cheng

Unsupervised Information Refinement Training of Large Language Models for Retrieval-Augmented Generation (2024.acl-long)

Challenge:	Existing approaches to tool learning rely on hand-crafted prompts and natural language reasoning, making multi-step planning difficult and lacking precise error diagnosis and reflection mechanisms.
Approach:	They propose a framework that reformulates tool learning as a code generation task.
Outcome:	The proposed framework achieves superior performance in task completion accuracy and execution reliability compared to existing approaches.

Copied to clipboard

Shicheng Xu, Liang Pang, Mo Yu, Fandong Meng, Huawei Shen, Xueqi Cheng, Jie Zhou

A Study of Implicit Ranking Unfairness in Large Language Models (2024.findings-emnlp)

Challenge:	Existing studies show that LLMs face challenges in effectively using retrieved information . authors propose a method that considers LLM as "Information Refiner"
Approach:	They propose a method that considers LLMs as "Information Refiners" they propose INFO-RAG, which is low-cost and general across various tasks .
Outcome:	The proposed method improves performance of LLaMA2 by 9.39% relative points . it is low-cost and general across various tasks, and is robust and in-context learning is possible .

Copied to clipboard

Chen Xu, Wenjie Wang, Yuxin Li, Liang Pang, Jun Xu, Tat-Seng Chua

Qsnail: A Questionnaire Dataset for Sequential Question Generation (2024.lrec-main)

Challenge:	Large language models (LLMs) have demonstrated superior ability to serve as ranking models, but they will exhibit discriminatory ranking behaviors based on users’ sensitive attributes (gender).
Approach:	They propose an evaluation method to investigate the severity of implicit ranking unfairness and a pair-wise regression method to conduct fair-aware data augmentation for LLM fine-tuning.
Outcome:	The proposed method outperforms existing methods in ranking fairness, achieving this with only a small reduction in accuracy.

Copied to clipboard

Yan Lei, Liang Pang, Yuanzhuo Wang, Huawei Shen, Xueqi Cheng

ReCoSa: Detecting the Relevant Contexts with Self-Attention for Multi-turn Dialogue Generation (P19-1)

Challenge:	Questionnaires are a professional research methodology used for qualitative and quantitative analysis of human opinions, preferences, and behaviors.
Approach:	They propose a questionnaire-based dataset that consists of 13,168 human-written questionnaires.
Outcome:	The proposed dataset contains 13,168 human-written questionnaires gathered from online platforms.

Copied to clipboard

Hainan Zhang, Yanyan Lan, Liang Pang, Jiafeng Guo, Xueqi Cheng

Dialogue Systems & Conversational Agents

Challenge:	Existing hierarchical recurrent encoder-decoder models treat all contexts indiscriminately, which may hurt the following response generation process.
Approach:	They propose a hierarchical recurrent encoder-decoder model that treats all contexts indiscriminately and uses a word level LSTM encoder to obtain the initial representation of each context.
Outcome:	The proposed model outperforms baseline models on Chinese customer services and English Ubuntu dialogue datasets in terms of both metric-based and human evaluations.

Subjective Topic meets LLMs: Unleashing Comprehensive, Reflective and Creative Thinking through the Negation of Negation (2024.emnlp-main)

Copied to clipboard

Fangrui Lv, Kaixiong Gong, Jian Liang, Xinyu Pang, Changshui Zhang

Reasoning

Challenge:	Large language models (LLMs) exhibit powerful reasoning capacity, but their evaluation still lacks comprehensiveness.
Approach:	They propose a framework grounded in the principle of the Negation of Negation (NeoN) to unleash the potential comprehensive, reflective, and creative thinking abilities of LLMs.
Outcome:	The proposed framework unleashes the potential comprehensive, reflective, and creative thinking abilities of large language models.

Enhancing Training Data Attribution for Large Language Models with Fitting Error Consideration (2024.emnlp-main)

Copied to clipboard

Kangxi Wu, Liang Pang, Huawei Shen, Xueqi Cheng

Do LLMs Play Dice? Exploring Probability Distribution Sampling in Large Language Models for Behavioral Simulation (2025.coling-main)

Challenge:	Large language models (LLMs) are difficult to interpret due to their black-box nature and randomness.
Approach:	They propose a new method which enhances influence functions by addressing fitting errors by eliminating knowledge bias present in the base model before fine-tuning.
Outcome:	The proposed method outperforms existing methods and achieves an average AUC of 91.64%.

Copied to clipboard

Jia Gu, Liang Pang, Huawei Shen, Xueqi Cheng

Self-Generated Critiques Boost Reward Modeling for Language Models (2025.naacl-long)

Challenge:	LLMs are used to emulate sequential decision-making processes of humans . however, their ability to perform probabilistic sampling is limited .
Approach:	They propose to use large language models (LLMs) as agents to emulate the sequential decision-making processes of humans represented as Markov decision-makers (MDPs).
Outcome:	The proposed models can understand probabilities, but struggle with sampling precision . integrating coding tools can improve sampling precision, but this level of sampling precision still makes it difficult to simulate human behavior as agents.

Copied to clipboard

Yue Yu, Zhengxing Chen, Aston Zhang, Liang Tan, Chenguang Zhu, Richard Yuanzhe Pang, Yundi Qian, Xuewei Wang, Suchin Gururangan, Chao Zhang, Melanie Kambadur, Dhruv Mahajan, Rui Hou

Stable-RAG: Mitigating Retrieval-Permutation-Induced Hallucinations in Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing reward models produce scalar scores and struggle to incorporate critiques in a natural language format.
Approach:	They propose a framework that predicts critiques and rewards using self-generated critiques without extra supervision.
Outcome:	The proposed framework improves reward modeling accuracy by 3.7%-7.3% compared to standard reward models and LLM judges.

Copied to clipboard

Qianchi Zhang, Hainan Zhang, Liang Pang, Hong-Wei Zheng, Zhiming Zheng

Responsible & Trustworthy NLP

Challenge:	Existing RAG methods focus on enhancing LLM robustness to low-quality retrieval, but neither address permutation sensitivity.
Approach:	They propose a method that exploits permutation sensitivity to mitigate hallucinations in Large Language Models.
Outcome:	The proposed model improves answer accuracy, reasoning consistency, and generalization across datasets, retrievers, and input lengths compared with strong baselines.

LLMDet: A Third Party Large Language Models Generated Text Detection Tool (2023.findings-emnlp)

Copied to clipboard

Kangxi Wu, Liang Pang, Huawei Shen, Xueqi Cheng, Tat-Seng Chua

MacLaSa: Multi-Aspect Controllable Text Generation via Efficient Sampling from Compact Latent Space (2023.findings-emnlp)

Challenge:	Existing detection tools rely on access to LLMs and can only distinguish between machine-generated and human-authored text.
Approach:	They propose a model-specific, secure, efficient, and extendable detection tool that can source text from specific LLMs.
Outcome:	The proposed tool can source text from specific LLMs, such as GPT-2, OPT, LLaMA, and others.

Copied to clipboard

Hanxing Ding, Liang Pang, Zihao Wei, Huawei Shen, Xueqi Cheng, Tat-Seng Chua

Green & Sustainable NLP

Challenge:	Existing approaches to multi-aspect controllable text generation require expensive iteration / searching within the discrete text space during the decoding stage, resulting in a degradation of text quality due to the domain discrepancies between different aspects.
Approach:	They propose a framework that estimates compact latent space for multiple aspects and performs efficient Sampling with a fast sampler to eliminate domain discrepancies.
Outcome:	The proposed framework outperforms baselines on attribute relevance and textual quality while maintaining a high inference speed.

SimOAP: Improve Coherence and Consistency in Persona-based Dialogue Generation via Over-sampling and Post-evaluation (2023.acl-long)

Copied to clipboard

Junkai Zhou, Liang Pang, Huawei Shen, Xueqi Cheng

Dialogue Systems & Conversational Agents

Challenge:	Existing work on large-scale corpora-based language models is limited and hard to generalize to all types of pre-trained language models.
Approach:	They propose a two-stage SimOAP strategy that over-samples and post-evaluates large-scale responses from existing models and selects a good response based on multiple evaluation metrics.
Outcome:	The proposed strategy outperforms baseline and automatic evaluation strategies in both automatic and human evaluations.

Projecting Out the Malice: A Global Subspace Approach to LLM Detoxification (2026.acl-long)

Copied to clipboard

Zenghao Duan, Zhiyi Yin, Zhichao Shi, Liang Pang, Shaoling Jing, Zihe Huang, Jiayi Wu, Yu Yan, Jingcheng Deng, Huawei Shen, Xueqi Cheng

Challenge:	Large language models (LLMs) exhibit exceptional performance but pose inherent risks of generating toxic content.
Approach:	They propose a method that removes toxic subspaces from FFN parameters . they propose to use a lightweight method to eliminate toxic subespaces .
Outcome:	The proposed method achieves SOTA detoxification while preserving general capabilities without large-scale retraining.

Optimal Partial Transport Based Sentence Selection for Long-form Document Matching (2022.coling-1)

Copied to clipboard

Weijie Yu, Liang Pang, Jun Xu, Bing Su, Zhenhua Dong, Ji-Rong Wen

Challenge:	Existing methods for document matching are limited by the partial nature of the sentence-level matching signals.
Approach:	They propose a matching approach that equips existing document matching models with an Optimal Partial Transport component, namely OPT-Match, which selects the key sentences that play a major role in matching.
Outcome:	The proposed approach outperforms existing models on four publicly available datasets and the key sentences selected by it are consistent with human-provided rationales.

METNet: A Mutual Enhanced Transformation Network for Aspect-based Sentiment Analysis (2020.coling-main)

Copied to clipboard

Bin Jiang, Jing Hou, Wanyue Zhou, Chao Yang, Shihan Wang, Liang Pang

Sentiment Analysis

Challenge:	Existing methods for learning complex sentences with multiple aspects are ill-equipped to learn complex sentences .
Approach:	They propose a mutual enhanced transformation network for the ABSA task . it improves representation learning of the aspect with contextual semantic features .
Outcome:	The proposed model improves representation learning of the aspect with contextual semantic features, giving the aspect more abundant information.

Beyond Language: Learning Commonsense from Images for Reasoning (2020.findings-emnlp)

Copied to clipboard

Wanqing Cui, Yanyan Lan, Liang Pang, Jiafeng Guo, Xueqi Cheng

Reasoning

Challenge:	Existing commonsense reasoning methods use raw texts to perform data representation and answer prediction tasks.
Approach:	They propose a novel approach to learn commonsense from images instead of limited raw texts or costly knowledge bases.
Outcome:	The proposed approach outperforms language-based methods on commonsense reasoning problems on two commonsence reasoning problems.

PEDNet: A Persona Enhanced Dual Alternating Learning Network for Conversational Response Generation (2020.coling-main)

Copied to clipboard

Bin Jiang, Wanyue Zhou, Jingxu Yang, Chao Yang, Shihan Wang, Liang Pang

Dialogue Systems & Conversational Agents

Challenge:	Existing persona-based dialogue models generate personalized responses using predefined persona information, but they lack personality.
Approach:	They propose a persona-based dual Alternating Learning Network that generates personalized responses using predefined persona information.
Outcome:	The proposed method produces more personalized responses than baseline methods.

The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis (2026.acl-long)

Copied to clipboard

Zihao Wei, Liang Pang, Jiahao Liu, Wenjie Shi, Jingcheng Deng, Shicheng Xu, Zenghao Duan, Jingang Wang, Fei Sun, Huawei Shen, Xueqi Cheng

Reasoning

Challenge:	Explicit reasoning trajectories increase performance but often trigger overthinking . despite its importance, this study examines how each step of reasoning affects the final outcome .
Approach:	They propose a Reasoning Completion Point Detector that detects the RCP by monitoring rank dynamics of termination tokens.
Outcome:	The proposed method reduces token usage by up to 44% while preserving accuracy.

Adaptive Information Seeking for Open-Domain Question Answering (2021.emnlp-main)

Copied to clipboard

Yunchang Zhu, Liang Pang, Yanyan Lan, Huawei Shen, Xueqi Cheng

Natural Language Interfaces

Challenge:	Existing iterative approaches to open-domain question answering use predefined strategies . e.g., BM25, DPR, and hyperlink are defined as actions .
Approach:	They propose a novel adaptive information-seeking strategy for open-domain question answering . they propose to use a partially observed Markov decision process to select a proper retrieval action .
Outcome:	Experiments on SQuAD Open and HotpotQA fullwiki show that AISO outperforms baseline methods with predefined strategies in retrieval and answer evaluations.

Following the Autoregressive Nature of LLM Embeddings via Compression and Alignment (2025.emnlp-main)

Copied to clipboard

Jingcheng Deng, Zhongtao Jiang, Liang Pang, Zihao Wei, Liwei Chen, Kun Xu, Yang Song, Huawei Shen, Xueqi Cheng

Physics Reasoner: Knowledge-Augmented Reasoning for Solving Physics Problems with Large Language Models (2025.coling-main)

Challenge:	Experimental results demonstrate that our method significantly outperforms traditional contrastive learning approaches when using the same amount of data.
Approach:	They propose a new contrastive learning method built on embedding conditional probability distributions that integrates two tasks: information compression and conditional distribution alignment.
Outcome:	The proposed method outperforms traditional contrastive learning approaches and achieves comparable performance to state-of-the-art models when using the same amount of data.

Copied to clipboard

Xinyu Pang, Ruixin Hong, Zhanke Zhou, Fangrui Lv, Xinwei Yang, Zhilong Liang, Bo Han, Changshui Zhang

Plot Retrieval as an Assessment of Abstract Semantic Association (2024.acl-srw)

Challenge:	Existing large language models (LLMs) fail due to lack of knowledge or incorrect knowledge application.
Approach:	They propose a knowledge-augmented framework that constructs a formula set to provide explicit physics knowledge and utilizes checklists to guide effective knowledge application.
Outcome:	The proposed framework achieves state-of-the-art performance on SciBench with an average accuracy improvement of 5.8%.

Copied to clipboard

Shicheng Xu, Liang Pang, Jiangnan Li, Mo Yu, Fandong Meng, Huawei Shen, Xueqi Cheng, Jie Zhou

Multi-level Adaptive Contrastive Learning for Knowledge Internalization in Dialogue Generation (2023.emnlp-main)

Challenge:	Existing information retrieval datasets cannot capture abstract semantic associations well.
Approach:	They propose a task that retrieves relevant plots from the book for a query using a labeled dataset.
Outcome:	The proposed task can be used to evaluate the performance of IR models on the novel task Plot Retrieval.

Copied to clipboard

Chenxu Yang, Zheng Lin, Lanrui Wang, Chong Tian, Liang Pang, Jiangnan Li, Qirong Ho, Yanan Cao, Weiping Wang

RegaVAE: A Retrieval-Augmented Gaussian Mixture Variational Auto-Encoder for Language Modeling (2023.findings-emnlp)

Challenge:	Existing knowledge-grounded dialogue generation models struggle with dull and repetitive outputs, a problem commonly termed as text degeneration.
Approach:	They propose a framework that allows the model to "cheat" the objective by duplicating knowledge segments in a superficial pattern matching based on overlap.
Outcome:	The proposed framework can be applied to a WoW dataset and shows that it works across models and decoding strategies.

Copied to clipboard

Jingcheng Deng, Liang Pang, Huawei Shen, Xueqi Cheng