Reproduce “Explainable Reasoning over Knowledge Graphs for Recommendation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiang Wang, Dingxian Wang, Canran Xu, Xiangnan He, Yixin Cao, Tat-Seng Chua
A model called Knowledge-aware Path Recurrent Network (KPRN) was developed to provide explainable recommendations by explicitly reasoning over multi-hop paths in knowledge graphs. It achieved superior recommendation performance on diverse datasets and generated interpretable, path-wise explanations for user preferences.
Key insights
- Explicitly modeling the sequential composition of entities and relations in multi-hop paths within a knowledge graph is more effective for inferring user preferences than implicit embedding regularization.
- The inclusion of explicit relation embeddings in path representations is crucial for capturing fine-grained semantic intent and distinguishing nuanced connections.
- A weighted pooling mechanism that allows different paths to contribute proportionally to the final recommendation score enhances both performance and the inherent explainability of the model.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.