Hierarchical attentive knowledge graph embedding for personalized recommendation

Knowledge graphs (KGs) have proven to be effective for high-quality recommendation, where the connectivities between users and items provide rich complementary information user-item interactions. Most existing methods, however, are insufficient exploit KGs capturing user preferences, as they either represent via paths with limited expressiveness or implicitly model them by propagating over entire KG inevitable noise. In this paper, we design a novel hierarchical attentive knowledge graph embedding (HAKG) framework recommendation. Specifically, HAKG first extracts expressive subgraphs that link pairs characterize their connectivities, which accommodate both semantics topology of KGs. The then encoded subgraph encoding generate embeddings enhanced preference prediction. Extensive experiments show superiority against state-of-the-art recommendation well its potential in alleviating data sparsity issue.