Semantic and Structural Clustering for Enhancing Subgraph-Based Knowledge Graph Completion

You-Syuan Chen, Cheng-Ru Wei, Tzi-Ping Gau, Wei-Po Lee

Abstract: Knowledge graphs play a key role in medical studies, aiding in disease diagnosis, understanding disease relationships, and enhancing knowledge integration. In this work, we employ structural and semantic-based methods for knowledge graph embedding and subgraph extraction using vector similarity. We also explore topic modeling for clustering. Our study compares different embedding techniques and clustering methods, confirming the effectiveness of our approach in link prediction tasks.