2026–Present
LLMs and Graph Algorithms
Exploring how large language models and LLM-based agents work with graph-structured information, particularly in tasks that require retrieving the right subgraph, following multi-step paths, or verifying evidence.
The current stage focuses on benchmark and literature analysis, including comparisons of GraCoRe, GraphSilo, GraphPile, and G-REAL.
Graph search
Subgraph retrieval
Path reasoning
Graph optimization
Evidence verification