SOOCHOW UNIVERSITY · NLP LAB
SUGAR
Soochow University Group of Advanced Retrieval (SUGAR)
Search, memory & reasoning for language AI.
SUGAR is an NLP research group within the NLP Lab at the School of Computer Science and Technology, Soochow University. We study how language models retrieve evidence, retain useful knowledge, and reason over it, with applications in conversational search and legal AI.
Led by Minghan Li, Associate Professor at Soochow University.
RESEARCH DIRECTIONS
From finding evidence to reasoning with it.
Search & RAG
Conversational search, long-document retrieval, query expansion, and evidence-aware reranking.
Agent memory & reasoning
Long-term conversational memory, causality-enhanced memory mechanisms, and efficient LLM reasoning.
Legal AI
Legal case retrieval, legal inference, and evidence ranking for knowledge-intensive tasks.
PEOPLE
Meet the people behind SUGAR.
Our group brings together researchers working on retrieval, language models, memory, reasoning, and legal AI.
Faculty
Minghan Li
Associate Professor · Group Lead
Master's Students · Year 3
Chao Zhang
Master's Student · Year 3
Master's Students · Year 2
Suchao An
Master's Student · Year 2
Tongna Chen
Master's Student · Year 2
Tianrui Lv
Master's Student · Year 2
Xinxuan Lv
Master's Student · Year 2
Co-supervised with Prof. Guodong Zhou
Yishuai Zhang
Master's Student · Year 2
Siqi Zhao
Master's Student · Year 2
Junjie Zou
Master's Student · Year 2
Co-supervised with Prof. Guodong Zhou
Master's Students · Year 1
Xiaoping Chen
Master's Student · Year 1
Luyao Li
Master's Student · Year 1
Jifeng Sun
Master's Student · Year 1
Hongshuo Xiao
Master's Student · Year 1
Co-supervised with Prof. Guodong Zhou
Qingyu Zhou
Master's Student · Year 1
Tong Zhu
Master's Student · Year 1
Undergraduate Students
Coming soon.
RECENT UPDATES
News from our research.
Our survey, A Survey of Long-Document Retrieval in the PLM and LLM Era, has been accepted by ACM TOIS.
Three papers accepted by EMNLP 2026: SAGA and Cascade-SC to the Main Conference, and EviRerank to Findings.
Retrieval-Feedback Aligned Distillation for Deployable LLM Query Expansion has been accepted for an oral presentation at NLPCC 2026.
Query Expansion in the Age of Pre-trained and Large Language Models: A Comprehensive Survey has been accepted by ACM TOIS.
PAPERS & PROJECTS
Explore the work behind the ideas.
Our publications span retrieval, query expansion, legal inference, and efficient reasoning. Browse representative work and the full publication list on Minghan Li’s academic homepage.