Papers by Qi Su

Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models (2026.findings-acl)

Challenge:	Existing studies on forex prediction ignore related text completely and focus on forex trade data only, which loses important semantic information.
Approach:	They propose a BERT-based Hierarchical Aggregation Model to summarize forex news . they group news from different aspects and extract the most crucial news in each group .
Outcome:	The proposed model outperforms baseline methods and grouping methods and summarizes the influence patterns for forex trading.

Copied to clipboard

Hengyuan Zhang, Zhihao Zhang, Ercong Nie, Mingyang Wang, Zunhai Su, Yiwei Wang, Qianli Wang, Shuzhou Yuan, Xufeng Duan, Qibo Xue, Zeping Yu, Chenming Shang, Xiao Liang, Jing Xiong, Hui Shen, Chaofan Tao, Zhengwu Liu, Senjie Jin, Zhiheng Xi, Dongdong Zhang, Sophia Ananiadou, Tao Gui, Ruobing Xie, Hayden Kwok-Hay So, Hinrich Schuetze, Xuanjing Huang, Qi Zhang, Ngai Wong

Explainability & Interpretability in NLP

Challenge:	Existing literature on mechanistic interpretation (MI) treats it as an observational science, leaving practical applications underexplored.
Approach:	They propose a survey structured around the pipeline to identify and improve MI models.
Outcome:	The proposed framework enables tangible improvements in Alignment, Capability, and Efficiency.

CHisIEC: An Information Extraction Corpus for Ancient Chinese History (2024.lrec-main)

Copied to clipboard

Xuemei Tang, Qi Su, Jun Wang, Zekun Deng

PQR: Improving Dense Retrieval via Potential Query Modeling (2025.acl-long)

Challenge:	Historical and cultural heritage preservation is an important branch of digital humanities, where the rich tapestry of the past meets the cutting-edge tools of the digital age.
Approach:	They present a dataset to evaluate NER and RE tasks in ancient Chinese history . they use four distinct entity types and twelve relation types to identify them .
Outcome:	The "Chinese Historical Information Extraction Corpus" is a dataset from 13 dynasties spanning over 1830 years . the dataset encompasses four distinct entity types and twelve relation types .

Copied to clipboard

Junfeng Kang, Rui Li, Qi Liu, Yanjiang Chen, Zheng Zhang, Junzhe Jiang, Heng Yu, Yu Su

Information Retrieval

Challenge:	Existing training data is sparse, with each document associated with one or a few labeled queries.
Approach:	They propose a training-free potential query retrieval framework to address this problem . they use a Gaussian mixture distribution to model all potential queries for a document .
Outcome:	The proposed method is able to capture comprehensive semantic information from a document with multiple queries.

Walking in Others’ Shoes: How Perspective-Taking Guides Large Language Models in Reducing Toxicity and Bias (2024.emnlp-main)

Copied to clipboard

Rongwu Xu, Zian Zhou, Tianwei Zhang, Zehan Qi, Su Yao, Ke Xu, Wei Xu, Han Qiu

Deconvolution-Based Global Decoding for Neural Machine Translation (C18-1)

Challenge:	Existing prompting methods that require white-box access to the model or substantial training fail to simultaneously lessen toxicity and bias.
Approach:	They propose a strategy that encourages LLMs to integrate diverse human perspectives and self-regulate their responses by incorporating diverse human viewpoints.
Outcome:	The proposed approach can significantly diminish toxicity (up to 89%) and bias (up 73%) in LLMs’ responses.

Copied to clipboard

Junyang Lin, Xu Sun, Xuancheng Ren, Shuming Ma, Jinsong Su, Qi Su

Machine Translation

Challenge:	Existing models for Neural Machine Translation (NMT) use Recurrent Neural Network (RNN) to generate translation word by word following a sequential order.
Approach:	They propose a Neural Machine Translation (NMT) model that decodes the sequence with the guidance of its structural prediction of the target-side context.
Outcome:	The proposed model is more competitive compared with the state-of-the-art methods and reduces repetition with the instruction from the target-side context for decoding.

ShopSimulator: Evaluating and Exploring RL-Driven LLM Agent for Shopping Assistants (2026.acl-long)

Copied to clipboard

Pei Wang, Yanan Wu, Xiaoshuai Song, Weixun Wang, Gengru Chen, Zhongwen Li, Kezhong Yan, Qi Liu, Ken Deng, Shuaibing Zhao, Shaopan Xiong, Xuepeng Liu, Xuefeng Chen, Wanxi Deng, Wenbo Su, Bo Zheng

Challenge:	Existing studies on large language model-based agents focus on evaluation benchmarks without training support.
Approach:	They propose a large-scale Chinese shopping simulation environment that uses large language models to train agents.
Outcome:	The proposed model performs poorly in a large-scale and challenging shopping environment in China.

UCS: Estimating Unseen Coverage for Improved In-Context Learning (2026.findings-acl)

Copied to clipboard

Jiayi Xin, Xiang Li, Evan Qiang, Weiqing He, Tianqi Shang, Weijie J Su, Qi Long

Text Clustering

Challenge:	Existing selection methods prioritize heuristic notions of relevance or diversity and provide limited insight into the coverage of a demonstration set.
Approach:	They propose a training-free, subset-level coverage prior that is unrevealed by a model-consistent embedding and a Smoothed Good-Turing estimator to estimate the number of unrevelled clusters within a candidate subset.
Outcome:	Experiments on multiple intent-classification and reasoning benchmarks show that augmenting strong baselines with UCS improves ICL accuracy by 2-6% under the same selection budget.

Unsupervised Chinese Word Segmentation with BERT Oriented Probing and Transformation (2022.findings-acl)

Copied to clipboard

Wei Li, Yuhan Song, Qi Su, Yanqiu Shao

Low-Resource NLP

Challenge:	Existing methods for unsupervised Chinese word segmentation exploit shallow semantic information, which can miss important context.
Approach:	They propose to take advantage of deep contextual semantic information with a self-training manner to transform it into explicit word segmentation ability.
Outcome:	The proposed approach achieves state-of-the-art F1 score on two CWS benchmark datasets.

Restoring Ancient Ideograph: A Multimodal Multitask Neural Network Approach (2024.lrec-main)

Copied to clipboard

Siyu Duan, Jun Wang, Qi Su

Multimodality

Challenge:	despite efforts to preserve cultural relics, many ancient artefacts have fallen prey to ravages of time, natural deterioration, or deliberate human actions.
Approach:	They propose a multimodal multitask restoration model that uses visual and context understanding to restore ancient texts.
Outcome:	The proposed model predicts damaged characters and generates restored images simultaneously.

Many Heads Are Better Than One: Improved Scientific Idea Generation by A LLM-Based Multi-Agent System (2025.acl-long)

Copied to clipboard

Haoyang Su, Renqi Chen, Shixiang Tang, Zhenfei Yin, Xinzhe Zheng, Jinzhe Li, Biqing Qi, Qi Wu, Hui Li, Wanli Ouyang, Philip Torr, Bowen Zhou, Nanqing Dong

Alleviating Exposure Bias via Multi-level Contrastive Learning and Deviation Simulation in Abstractive Summarization (2023.findings-acl)

Challenge:	Recent AI methods have shown promise in tasks such as hypothesis generation and experimental design, but they fail to replicate the collaborative nature of real-world scientific practices.
Approach:	They propose a virtual scientific system that mimics the collaborative nature of scientific research by organizing a team of agents to generate, evaluate, and refine research ideas.
Outcome:	The proposed system outperforms the state-of-the-art method in producing new scientific ideas and offers valuable insights to guide future research.

Copied to clipboard

Jiawen Xie, Qi Su, Shaoting Zhang, Xiaofan Zhang

RePair: Automated Program Repair with Process-based Feedback (2024.findings-acl)

Challenge:	Abstractive summarization systems have a severe mismatch between training and inference, i.e., exposure bias.
Approach:	They propose a multi-level contrastive learning framework for abstractive summarization and a tailored sparse decoder self-attention pattern to bridge the gap between training and inference.
Outcome:	The proposed framework outperforms the state-of-the-art models on two summarization datasets while adding relatively low overhead.

Copied to clipboard

Yuze Zhao, Zhenya Huang, Yixiao Ma, Rui Li, Kai Zhang, Hao Jiang, Qi Liu, Linbo Zhu, Yu Su

Structure Regularized Neural Network for Entity Relation Classification for Chinese Literature Text (N18-2)

Challenge:	Commercial-scale language models (LMs) have taken APR to unprecedented levels, but they are limited by parameters and humans interact with them through explicit prompts.
Approach:	They propose a method that utilizes process supervision to improve program repair by allowing users to input feedback from compilers and test cases.
Outcome:	The proposed method outperforms large outcome-based generation methods and is inspired by strategies used in programming competitions.

Copied to clipboard

Ji Wen, Xu Sun, Xuancheng Ren, Qi Su

A Global Past-Future Early Exit Method for Accelerating Inference of Pre-trained Language Models (2021.naacl-main)

Challenge:	Existing methods for relation classification have been used in natural language processing.
Approach:	They propose a relation classification task for Chinese literature text using a new dataset.
Outcome:	The proposed model outperforms the state-of-the-art methods on Chinese literature text.

Copied to clipboard

Kaiyuan Liao, Yi Zhang, Xuancheng Ren, Qi Su, Xu Sun, Bin He

Dim-Krum: Backdoor-Resistant Federated Learning for NLP with Dimension-wise Krum-Based Aggregation (2022.findings-emnlp)

Challenge:	Existing methods to accelerate inference speed of pre-trained language models are limited to local representations of exit layer . current models are associated with large memory requirement and high computational cost, which slow down inference and further encumber the application of PLMs.
Approach:	They propose a method to exit early without passing through all inference layers . they take into consideration all the linguistic information embedded in the past layers a global perspective .
Outcome:	The proposed method outperforms existing methods by a large margin . it uses linguistic information embedded in the past layers and future features . the proposed method is scalable and cost-effective .

Copied to clipboard

Zhiyuan Zhang, Qi Su, Xu Sun

Robustness in NLP

Challenge:	Existing methods to reduce backdoor risk of federated learning are validated in the CV field.
Approach:	They propose a federated aggregation algorithm that detects errors determined by backdoor strengths for NLP attacks.
Outcome:	The proposed method is hard to defend against than CV, and the results validate it.

Global Encoding for Abstractive Summarization (P18-2)

Copied to clipboard

Junyang Lin, Xu Sun, Shuming Ma, Qi Su

Text Generation

Challenge:	Existing models for abstractive summarization suffer from repetition and semantic irrelevance.
Approach:	They propose a global encoding framework which controls the information flow from the encoder to the decoder based on the global information of the source context.
Outcome:	The proposed model outperforms baseline models on the LCSTS and English Gigaword and can generate summary of higher quality and reduce repetition.

DiffuVST: Narrating Fictional Scenes with Global-History-Guided Denoising Models (2023.findings-emnlp)

Copied to clipboard

Shengguang Wu, Mei Yuan, Qi Su

Multimodality

Challenge:	Existing methods for visual storytelling suffer from low inference speed and are not well-suited for synthetic scenes.
Approach:	They propose a diffusion-based system that generates visual descriptions as a single conditional denoising process.
Outcome:	The proposed system improves inter-sentence coherence and image-to-text fidelity.

Too Consistent to Detect: A Study of Self-Consistent Errors in LLMs (2025.emnlp-main)

Copied to clipboard

Hexiang Tan, Fei Sun, Sha Liu, Du Su, Qi Cao, Xin Chen, Jingang Wang, Xunliang Cai, Yuanzhuo Wang, Huawei Shen, Xueqi Cheng

Challenge:	Existing detection methods fail to account for self-consistent error . study identifies self-consistency errors and evaluates them .
Approach:	They propose a method that fuses hidden state evidence from an external verifier LLM to detect self-consistent errors.
Outcome:	The proposed method significantly enhances performance on self-consistent errors across three LLM families.

Specificity-Driven Cascading Approach for Unsupervised Sentiment Modification (D19-1)

Copied to clipboard

Pengcheng Yang, Junyang Lin, Jingjing Xu, Jun Xie, Qi Su, Xu Sun

Sentiment Analysis

Challenge:	Existing methods for unsupervised sentiment modification lack specific information in text generated without parallel data . specificity-driven cascading approach can improve specificity of generated text and content preservation .
Approach:	They propose a specificity-driven cascading approach for unsupervised sentiment modification . the method performs target sentiment addition and content reconstruction independently .
Outcome:	The proposed method outperforms competitive systems by a large margin on Yelp and Amazon datasets.

GA-SAM: Gradient-Strength based Adaptive Sharpness-Aware Minimization for Improved Generalization (2022.emnlp-main)

Copied to clipboard

Zhiyuan Zhang, Ruixuan Luo, Qi Su, Xu Sun

That Slepen Al the Nyght with Open Ye! Cross-era Sequence Segmentation with Switch-memory (2022.acl-long)

Challenge:	Recent studies show flat minima tend to imply better generalization abilities . however, it has some difficulty implying SAM to some natural language tasks .
Approach:	They propose a flatness-aware minimization algorithm that can be applied to natural language tasks . they propose to use parameter corruptions to explain why flat minima generalize better .
Outcome:	The proposed algorithm can generalize better for flat minima that are robust against corruptions or perturbations.

Copied to clipboard

Xuemei Tang, Qi Su

Syntactic Text Processing

Challenge:	Grammar, vocabulary, and lexical semantic shifts take place over time, creating diachronic linguistic gap.
Approach:	a Chinese cross-era learning framework incorporates era-specific linguistic knowledge into a neural network.
Outcome:	a new learning framework for Chinese word segmentation improves on four corpora from different eras.

AJ-Bench: Benchmarking Agent-as-a-Judge for Environment-Aware Evaluation (2026.findings-acl)

Copied to clipboard

Wentao Shi, Yu Wang, Yuyang Zhao, Yuxin Chen, Fuli Feng, Xueyuan Hao, Xi Su, Qi GU, Hui Su, Xunliang Cai, Xiangnan He

Challenge:	Existing approaches to verify agent behaviors in complex environments rely on rule-based verifiers or LLM-as-a-Judge models.
Approach:	They propose a benchmark to evaluate Agent-as-a-Judge across three domains . the benchmark covers search, data systems, and graphical user interfaces - with 155 tasks and 516 trajectories .
Outcome:	The proposed benchmark outperforms existing benchmarks in search, data systems, and GUI domains while revealing open challenges in agent-based verification.

The Subtle Art of Defection: Understanding Uncooperative Behaviors in LLM based Multi-Agent Systems (2026.eacl-industry)

Copied to clipboard

Devang Kulshreshtha, Wanyu Du, Raghav Jain, Srikanth Doss, Hang Su, Sandesh Swamy, Yanjun Qi

Challenge:	Existing literature on uncooperative behavior degrades collective outcomes and requires more resilient multi-agent systems.
Approach:	They propose a game theory-based taxonomy of uncooperative agent behaviors and a structured, multi-stage simulation pipeline that dynamically generates and refines uncooperation behaviors as agents’ states evolve.
Outcome:	The proposed framework achieves 96.7% accuracy in generating realistic uncooperative behaviors, validated by human evaluations.

Pretrain-KGE: Learning Knowledge Representation from Pretrained Language Models (2020.findings-emnlp)

Copied to clipboard

Zhiyuan Zhang, Xiaoqian Liu, Yi Zhang, Qi Su, Xu Sun, Bin He

Neural Network Surgery: Injecting Data Patterns into Pre-trained Models with Minimal Instance-wise Side Effects (2021.naacl-main)

Challenge:	Existing knowledge graph embedding models suffer from limited knowledge representation due to sparse and noisy dataset annotations.
Approach:	They propose to use pretrained language models to enhance knowledge representation by leveraging world knowledge from pretrained models.
Outcome:	Extensive experiments show that the proposed framework can improve results over existing models.

Copied to clipboard

Zhiyuan Zhang, Xuancheng Ren, Qi Su, Xu Sun, Bin He

Semantic-Unit-Based Dilated Convolution for Multi-Label Text Classification (D18-1)

Challenge:	Existing neural network tuning methods cause instance-wise side effects . et al., 2018: a new approach to perform neural network surgery .
Approach:	They propose to perform neural network surgery by only changing 10-5 parameters . they propose to use a dynamic selecting method to achieve the best overall performance .
Outcome:	The proposed method achieves the best overall performance and induces fewer instance-wise side effects by changing only 10-5 of the parameters.

Copied to clipboard

Junyang Lin, Qi Su, Pengcheng Yang, Shuming Ma, Xu Sun

UniRAG: Unified Query Understanding Method for Retrieval Augmented Generation (2025.acl-long)

Challenge:	a novel model for multi-label text classification is proposed for the task of assigning multiple labels for a given text.
Approach:	They propose a novel model for multi-label text classification based on sequence-to-sequence learning and a hybrid attention mechanism that extracts both the word-level and the semantic unit.
Outcome:	The proposed model is competitive to the baseline models and more robust to classifying low-frequency labels.

Copied to clipboard

Rui Li, Liyang He, Qi Liu, Zheng Zhang, Heng Yu, Yuyang Ye, Linbo Zhu, Yu Su

Information Retrieval

Challenge:	Existing query augmentation methods face knowledge update lag and hallucinations in large language models (LLMs) Existing methods face two key challenges: (1) separation of query augmented and encoding tasks, which hinders information sharing and introduces cumulative errors; (2) difficulty of selecting optimal augmentation strategy for different scenarios.
Approach:	They propose a unified framework for query understanding in RAG that integrates internal and external knowledge to enhance query augmentation and encoding tasks.
Outcome:	The proposed framework outperforms traditional query augmentation methods in five knowledge-intensive benchmark tasks in both closed and open domain question answering.

Low-Entropy Watermark Detection via Bayes’ Rule Derived Detector (2025.findings-acl)

Copied to clipboard

Beining Huang, Du Su, Fei Sun, Qi Cao, Huawei Shen, Xueqi Cheng

Numerical Reasoning

Challenge:	Existing methods for text watermarking ignore strong evidences embedded in low-entropy tokens, causing statistical measures to falsely indicate the absence of a watermark.
Approach:	They propose a Bayes' Rule derived watermark Detector which exploits watermark information from every token by leveraging the posterior probability of watermark’s presence.
Outcome:	The proposed method achieves 50% and 70% relative improvements over baselines in code generation and math problem-solving tasks.

metaCAT: A Metadata-based Task-oriented Chatbot Annotation Tool (2020.aacl-demo)

Copied to clipboard

Ximing Liu, Wei Xue, Qi Su, Weiran Nie, Wei Peng

Natural Language Interfaces

Challenge:	Creating high-quality annotated dialogue corpora necessitates a high level of human engagements.
Approach:	They propose to develop an annotation tool specifically for developing task-oriented dialogue data that provides comprehensive metadata annotation coverage to the domain, intent, and span information.
Outcome:	The tool provides comprehensive metadata annotation coverage to domain, intent, and span information.

Rethinking Pragmatics in Large Language Models: Towards Open-Ended Evaluation and Preference Tuning (2024.emnlp-main)

Copied to clipboard

Shengguang Wu, Shusheng Yang, Zhenglun Chen, Qi Su

Sina Mandarin Alphabetical Words:A Web-driven Code-mixing Lexical Resource (2020.aacl-main)

Challenge:	Existing methods to assess social-pragmatic inference in large language models are inadequacy, and preferential tuning is the best approach.
Approach:	They propose to use free-form models' responses as a measure to assess social-pragmatic reasoning and advocate for preference optimization over supervised finetuning (SFT).
Outcome:	The proposed model outperforms supervised finetuning (SFT) and offers a near-free launch in pragmatic abilities without compromising general capabilities.

Copied to clipboard

Rong Xiang, Mingyu Wan, Qi Su, Chu-Ren Huang, Qin Lu

Challenge:	Mandarin Alphabetical Words (MAWs) are a key component of Modern Chinese . they are characterized by unique code-mixing idiosyncrasies influenced by language exchanges .
Approach:	They propose to construct a large collection of Mandarin Alphabetic Words from Sina Weibo . they propose to use a web-based technique to identify and validate MAWs .
Outcome:	The proposed method identifies 16,207 Mandarin Alphabetic Words (MAWs) using a web-based technique . the results show that the proposed method is useful for linguistic research and inquiries .

Golden Touchstone: A Comprehensive Bilingual Benchmark for Evaluating Financial Large Language Models (2025.findings-emnlp)

Copied to clipboard

Xiaojun Wu, Junxi Liu, Huan-Yi Su, Zhouchi Lin, Yiyan Qi, Chengjin Xu, Jiajun Su, Jiajie Zhong, Fuwei Wang, Saizhuo Wang, Fengrui Hua, Jia Li, Jian Guo

An Effective Incorporating Heterogeneous Knowledge Curriculum Learning for Sequence Labeling (2025.acl-short)

Challenge:	Existing financial benchmarks suffer from limited language and task coverage, low-quality datasets, and inadequate adaptability for LLM evaluation.
Approach:	They propose a bilingual benchmark for financial LLMs that assesses models’ language understanding and generation capabilities.
Outcome:	The proposed bilingual benchmark assesses models’ language understanding and generation capabilities.

Copied to clipboard

Xuemei Tang, Jun Wang, Qi Su, Chu-Ren Huang, Jinghang Gu