Node-personalized multi-graph convolutional networks for recommendation

Tiantian Zhou, Hailiang Ye, Feilong Cao

Abstract: Highlights•A node-personalized multi-graph convolutional network is proposed for recommendations.•A compound-hop information propagation and aggregation framework is designed.•A node importance awareness block is devised.•A Jaccard similarity and co-occurrence matrix fusion graph construction module is built.