Graphs serve as a powerful abstraction for modeling relational structures in data and play a vital role in applications ranging from social networks to AI reasoning systems. In this talk, I will present a series of algorithmic and system-level advances at the intersection of graph mining and large language models (LLMs). I will begin with foundational algorithmic work on densest subgraph discovery and community search, including our recent results. I will then introduce GraphRAG, a retrieval-augmented generation framework that leverages graph-based knowledge to improve factuality and interpretability in LLM outputs. Finally, I will discuss our ongoing work on graph-based agents, which equip LLMs with structured memory, reasoning over subgraph states, and execution planning in multi-agent settings. These efforts collectively demonstrate how bridging graphs with LLMs can unlock new capabilities for data-intensive applications.
Short Bio: Yingli Zhou is a final-year Ph.D. candidate at The Chinese University of Hong Kong, Shenzhen, under the supervision of Prof. Yixiang Fang. His research lies at the intersection of graph data management and large language models. He has published 12 papers in top-tier conferences such as SIGMOD and VLDB, including 7 as the first author. His contributions span scalable algorithms for densest subgraph discovery, retrieval-augmented generation on graphs (GraphRAG), and LLM-based agent systems with structured memory. He has also collaborated with industry partners, including Huawei Cloud and Alibaba Tongyi Lab. Yingli will soon join the National University of Singapore as a visiting researcher under the supervision of Prof. Xiaokui Xiao.
