Cui, Limeng, and Dongwon Lee. "Ketch: Knowledge graph enhanced thread recommendation in healthcare forums." Proceedings of the 45th international acm sigir conference on research and development in information retrieval. 2022.

The paper introduces the KETCH model for enhancing thread recommendations in online healthcare forums. Unlike traditional methods that focus solely on post content, KETCH utilizes a knowledge graph to incorporate user symptoms and clinical conditions, improving the relevance of recommended threads. The model leverages graph neural networks to represent interactions among users, threads, and posts, and includes a knowledge-enhanced attention mechanism to better capture latent conditions and symptoms. Applied to drug side effect prediction, KETCH outperforms seven existing methods on multiple datasets. The source code for KETCH is available at GitHub.