Clara (Chen) Shao

LLM-GALP (Large Language Model-featured Graph Algorithms for Link Prediction) is a research project I lead as principal investigator, funded by the German Federal Ministry of Education and Research (BMFTR), 2025–2026 (EUR 105,800).

The project develops memory-efficient methods for fine-tuning large language models on text-attributed graphs, combining the semantic power of language models with the structural inductive biases of graph neural networks. As part of this work, we introduce a new benchmark for link prediction on text-attributed graphs and an efficient LM-nested GNN design, and study how edge-level automorphism affects link prediction performance.

Related work: Bridging Semantics and Structure: A Theoretically Grounded Framework for Link Prediction on TAGs (TKDD 2026) and Edge-Level Automorphism in GNNs (preprint).

This project features the excellent work of supervised student Ruirui Wang.