Papers by Wenjun Liu
EPIR: Capturing Promoting and Inhibiting Relationships between Events (2026.findings-acl)
Copied to clipboard
| Challenge: | promoting and inhibiting relationships capture directional, probabilistic, and context-dependent shifts in event likelihood. |
| Approach: | They propose a framework for estimating promoting and inhibiting relationships from observed event data. |
| Outcome: | The proposed framework outperforms state-of-the-art models on real-world datasets in accuracy. |
From Outcome to Process: Optimizing MoE Load Balancing with MCTS (2026.findings-acl)
Copied to clipboard
Wenjun Ke, Hengyuan Xu, Ziyu Shang, Yao He, Jiahao Wang, Zijie Xu, Peng Wang, Yuhang Lou, Jiajun Liu
| Challenge: | Existing balancing strategies focus on constraining the final distribution of expert usage, but overlook the routing decisions made at each layer. |
| Approach: | They propose a three-stage framework that leverages process-level rewards to guide balanced expert routing. |
| Outcome: | Extensive experiments show that LayerMoE improves the performance of state-of-the-art LoRA-MoA baselines, yielding an average accuracy gain of 1.39%. |
LUNE: Efficient LLM Unlearning via LoRA Fine-Tuning with Negative Examples (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models encode vast factual knowledge, yet their inability to selectively forget specific information hinders privacy protection, bias mitigation, and post-deployment correction. |
| Approach: | They propose a LoRA-based negative-only unlearning framework that updates only low-rank adapters while freezing the backbone. |
| Outcome: | The proposed framework reduces computational cost by about an order of magnitude compared to full fine-tuning and memory-editing methods. |
GeoLaux: A Benchmark for Evaluating MLLMs’ Geometry Performance on Long-Step Problems Requiring Auxiliary Lines (2026.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks for Geometry problem solving lack fine-grained evaluation for long-step problems necessitating auxiliary line construction. |
| Approach: | They present a fine-grained annotated dataset with long-step reasoning and auxiliary line construction that provides a detailed evaluation of 23 leading MLLMs. |
| Outcome: | The proposed model performs significantly worse on long-step problems than short-step ones, with 18 models showing a performance drop of over 50%. |
Subtle Errors in Reasoning: Preference Learning via Error-injected Self-editing (2025.acl-long)
Copied to clipboard
Kaishuai Xu, Tiezheng Yu, Wenjun Hou, Yi Cheng, Chak Tou Leong, Liangyou Li, Xin Jiang, Lifeng Shang, Qun Liu, Wenjie Li
| Challenge: | Existing studies to improve mathematical ability typically involve applying preference learning to step-wise solution pairs, but they overlook critical subtle errors. |
| Approach: | They propose a preference learning framework that injects predefined subtle errors into pivotal tokens to construct hard pairs for error mitigation. |
| Outcome: | Extensive experiments show that the proposed framework improves on Qwen2-7B-Instruct and MATH with 4.5K training samples. |
On the Consistency of Commonsense in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of commonsense for large language models focus on downstream knowledge tasks, failing to probe whether LLMs truly understand and utilize knowledge or merely memorize it. |
| Approach: | They propose to automatically construct a large benchmark named CoCo which measures LLMs’ knowledge memorization, comprehension, and application and examines the consistency between these tasks. |
| Outcome: | The proposed benchmark systematically assesses LLMs’ knowledge memorization, comprehension, and application and examines the consistency between these tasks. |
Acquisition and Application of Novel Knowledge in Large Language Models (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for constructing new datasets rely on timestamps or simple templates that do not accurately reflect the real world. |
| Approach: | They propose a knowledge dataset construction approach that simulates biological evolution using knowledge graphs to generate synthetic entities with diverse attributes. |
| Outcome: | The proposed framework outperforms knowledge augmentation methods by 3.3%-38%. |
ORGAN: Observation-Guided Radiology Report Generation via Tree Reasoning (2023.acl-long)
Copied to clipboard
| Challenge: | Existing methods to generate radiology reports only rely on high-level plans, but they lack important information. |
| Approach: | They propose an Observation-guided radiology Report Generation framework which generates free-text descriptions for a set of radiographs. |
| Outcome: | The proposed framework outperforms state-of-the-art methods regarding text quality and clinical efficacy. |
Soft Knowledge Prompt: Help External Knowledge Become a Better Teacher to Instruct LLM in Knowledge-based VQA (2024.acl-long)
Copied to clipboard
| Challenge: | Recent research focuses on improving prediction performance and reliability of LLM. |
| Approach: | They propose a method to actively extract valuable information from the knowledge to produce a latent vector as a soft prompt, which is fused with the image embedding to form a knowledge-enhanced context to instruct LLM. |
| Outcome: | The proposed method improves performance on knowledge-based VQA benchmarks. |
RECAP: Towards Precise Radiology Report Generation via Dynamic Disease Progression Reasoning (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies have focused on producing concise observations while neglecting the precise attributes that determine the severity of diseases. |
| Approach: | They propose a model that generates precise radiology reports via dynamic disease progression reasoning by combining historical and spatiotemporal information. |
| Outcome: | Experiments on two publicly available datasets show the proposed model can generate precise and accurate radiology reports with dynamic disease progression reasoning. |
DS-MHP: Improving Chain-of-Thought through Dynamic Subgraph-Guided Multi-Hop Path (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing knowledge graph methods lack adaptability in knowledge-intensive tasks with multiple entities and implicit multi-hop relations. |
| Approach: | They propose a zero-shot framework to enhance LLM reasoning in multi-entity relation tasks. |
| Outcome: | DS-MHP outperforms baselines and state-of-the-art methods on 12 datasets spanning commonsense, logical, symbolic, and arithmetic reasoning. |
Diagram-Driven Course Questions Generation (2025.emnlp-main)
Copied to clipboard
Xinyu Zhang, Lingling Zhang, Yanrui Wu, Muye Huang, Wenjun Wu, Bo Li, Shaowei Wang, Basura Fernando, Jun Liu
| Challenge: | Visual Question Generation (VQG) research focuses on natural images while neglecting diagrams, a critical component of educational materials. |
| Approach: | They propose a diagram-driven course questions generation task to generate diagram-relevant questions for specific courses. |
| Outcome: | The proposed framework outperforms existing models on DiagramQG while maintaining strong generalizability across natural image datasets. |
Contrastive Pre-training for Personalized Expert Finding (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to expert finding are effective for a community question answering platform. |
| Approach: | They propose a CQA-domain Contrastive Pre-training framework for Expert Finding which could learn more comprehensive question representations. |
| Outcome: | The proposed framework could learn more comprehensive question representations on six real-world datasets. |
MultiConIR: Towards Multi-Condition Information Retrieval (2025.findings-emnlp)
Copied to clipboard
Xuan Lu, Sifan Liu, Bochao Yin, Yongqi Li, Xinghao Chen, Hui Su, Yaohui Jin, Wenjun Zeng, Xiaoyu Shen
| Challenge: | MultiConIR is a benchmark designed to evaluate retrieval and reranking models under nuanced multi-condition query scenarios. |
| Approach: | They propose a benchmark to evaluate retrieval and reranking models under nuanced multi-condition query scenarios. |
| Outcome: | The proposed benchmark evaluates retrieval and reranking models under nuanced multi-condition query scenarios across five domains. |
ODD: A Benchmark Dataset for the Natural Language Processing Based Opioid Related Aberrant Behavior Detection (2024.naacl-long)
Copied to clipboard
Sunjae Kwon, Xun Wang, Weisong Liu, Emily Druhl, Minhee Sung, Joel Reisman, Wenjun Li, Robert Kerns, William Becker, Hong Yu
| Challenge: | Opioid related aberrant behaviors (ORABs) present novel risk factors for opioid overdose. |
| Approach: | They propose to use a biomedical natural language processing benchmark dataset to classify ORABs from patients’ EHR notes into nine categories: confirmed aberrant behavior, suggested aberrant behaviors, Opioids, indication, diagnosed opioid dependency, Benzodiazepines, medication changes, and Central Nervous System-related. |
| Outcome: | The proposed dataset outperforms two state-of-the-art models in most categories and the gains are especially higher among uncommon classes. |
Eval-RAR: Evaluation-Driven Retrieval-Augmented Reasoning via Reinforcement Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for retrieval-augmented generation fail to provide explicit supervision for internal reasoning process. |
| Approach: | They propose an Evaluation-driven Retrieval-Augmented Reasoning framework that uses reinforcement learning and a fine-grained evaluation reward to optimize the process. |
| Outcome: | Eval-RAR outperforms existing methods on QA benchmarks on seven single-hop and multi-hop tasks. |
Capability Decomposition for Unified Information Extraction via Hierarchical Mixture-of-Experts (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for IE tasks suffer from inconsistent schema representation and implicitly intermediate reasoning . UC-UIE adopts a low-rank adapted hierarchical Mixture-of-Experts adapter for UIE tasks . |
| Approach: | They propose a framework that decomposes IE reasoning into three universal capabilities . UC-UIE adopts a low-rank Adaptation adapter to fine-tune LLMs for IE tasks . |
| Outcome: | The proposed framework outperforms full-parameter tuning methods with 1.24% trainable parameters and outperformed existing methods in generalization and interpretability. |
ICON: Improving Inter-Report Consistency in Radiology Report Generation via Lesion-aware Mixup Augmentation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to radiology report generation lack inter-report consistency, exhibiting biases towards common patterns and susceptibility to lesion variants. |
| Approach: | They propose a method which improves the inter-report consistency of radiology report generation by extracting lesions from input images and examining their characteristics. |
| Outcome: | The proposed system captures similarities in semantically equivalent lesions and can be used to generate reports for two semantically identical cases. |
LLM-Guided Semantic-Aware Clustering for Topic Modeling (2025.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that topic modeling is competitive compared to closed-source methods. |
| Approach: | They propose a semi-supervised topic modeling method that combines LLMs with clustering to improve topic generation and distribution. |
| Outcome: | The proposed method outperforms state-of-the-art methods that utilize GPT-4 on topic alignment and exhibits competitive performance compared to Neural Topic Models on topic quality. |
Boosting Textural NER with Synthetic Image and Instructive Alignment (2024.findings-acl)
Copied to clipboard
| Challenge: | Named entity recognition (NER) is a key task reliant on textual data. |
| Approach: | They propose a method to transform NER into a multimodal task by using images from the internet as auxiliaries. |
| Outcome: | The proposed method surpasses all text-only baselines and improves F1 score by 1.4% to 2.3% on prominent MNER datasets. |
Giving Control Back to Models: Enabling Offensive Language Detection Models to Autonomously Identify and Mitigate Biases (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models often rely on specific words to predict offensive content, compromising model fairness and potentially exacerbates biases against vulnerable and minority groups. |
| Approach: | They propose a bias self-awareness and data self-iteration framework to help models identify and mitigate biases by integrating multiple natural language processing techniques. |
| Outcome: | The proposed framework reduces false positive rate of models in in-distribution and out-of-difference tests, enhances model accuracy and fairness, and shows promising performance improvements on larger datasets. |
Judging with Many Minds: Do More Perspectives Mean Less Prejudice? On Bias Amplification and Resistance in Multi-Agent Based LLM-as-Judge (2025.findings-emnlp)
Copied to clipboard
Chiyu Ma, Enpei Zhang, Yilun Zhao, Wenjun Liu, Yaning Jia, Peijun Qing, Lin Shi, Arman Cohan, Yujun Yan, Soroush Vosoughi
| Challenge: | LLM-as-Judge frameworks provide scalable alternative to human evaluation . but the question of how intrinsic biases manifest in these settings remains unexplored . |
| Approach: | They conduct systematic analysis of four bias types in multi-agent LLM-as-Judge frameworks . they find debate framework amplifies biases sharply after initial debate . |
| Outcome: | The proposed frameworks amplify biases after debate and show they are stronger in meta-judge scenarios. |
Learning to Align Multi-Faceted Evaluation: A Unified and Robust Framework (2025.findings-acl)
Copied to clipboard
Kaishuai Xu, Tiezheng Yu, Yi Cheng, Wenjun Hou, Liangyou Li, Xin Jiang, Lifeng Shang, Qun Liu, Wenjie Li
| Challenge: | Existing methods for fine-tuning open-source LLMs are limited to text-based analysis under predefined general criteria. |
| Approach: | They propose a framework that fine-tunes LLMs to replicate the evaluation explanations and judgments of proprietary models. |
| Outcome: | The proposed evaluation framework outperforms existing fine-tuned evaluation methods in effectiveness and robustness. |
Adaptive Tool Use in Large Language Models with Meta-Cognition Trigger (2025.acl-long)
Copied to clipboard
Wenjun Li, Dexun Li, Kuicai Dong, Cong Zhang, Hao Zhang, Weiwen Liu, Yasheng Wang, Ruiming Tang, Yong Liu
| Challenge: | Existing research expands the tool arrays of large language models (LLMs), but the necessity of using these tools is often overlooked, leading to indiscriminate tool invocation. |
| Approach: | They propose a meta-cognition proxy proxy for LLMs self-assessment of their capabilities, reflecting the model’s awareness of its own limitations. |
| Outcome: | The proposed strategy is fine-tuned-free and costs minimal. |
Unlocking Instructive In-Context Learning with Tabular Prompting for Relational Triple Extraction (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for relational triple extraction (RTE) are unnatural and recast RTE tasks to text-to-text prompting formats. |
| Approach: | They propose a tabular prompting for RTE which frames RTE task into a table generation task and propose an instructive in-context learning which only selects and annotates samples considering triple semantics in massive unlabeled samples. |
| Outcome: | The proposed prompting for RTE with TableIE achieves state-of-the-art performance compared to other methods . the proposed prompts are based on off-the shelf LLMs and are scalable to multiple scenarios . |
Exploring Layer Activation Dynamic of CoT via Knowledge Probe (2026.acl-long)
Copied to clipboard
| Challenge: | Chain-of-thought reasoning has emerged as a crucial paradigm for multi-step reasoning tasks. |
| Approach: | They propose a multi-stage probing framework that enforces structured reasoning with three explicit stages: keyword extraction, theorem generation, and computation execution. |
| Outcome: | The proposed framework enforces structured reasoning with three explicit stages: keyword extraction, theorem generation, and computation execution. |
Pre-trained Personalized Review Summarization with Effective Salience Estimation (2023.findings-acl)
Copied to clipboard
| Challenge: | Pretrained language models (PLMs) are a new paradigm in text generation for the strong ability of natural language comprehension. |
| Approach: | They propose a pre-trained personalized review summarization method that incorporates personalized information into the salience estimation of input reviews. |
| Outcome: | The proposed method performs better than the state-of-the-art methods on real-world datasets. |
RADAR: Enhancing Radiology Report Generation with Supplementary Knowledge Injection (2025.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to enhance radiology report generation overlook the knowledge already embedded within the models, leading to redundant information integration. |
| Approach: | They propose a framework for enhancing radiology report generation with supplementary knowledge injection that leverages both internal and external knowledge. |
| Outcome: | Extensive experiments on MIMIC-CXR, CheXpert-Plus, and IU X-ray show that the proposed model outperforms state-of-the-art LLMs in both language quality and clinical accuracy. |