RippleNet

Abstract

To address the sparsity and cold start problem of collaborative filtering,\nresearchers usually make use of side information, such as social networks or\nitem attributes, to improve recommendation performance. This paper considers\nthe knowledge graph as the source of side information. To address the\nlimitations of existing embedding-based and path-based methods for\nknowledge-graph-aware recommendation, we propose Ripple Network, an end-to-end\nframework that naturally incorporates the knowledge graph into recommender\nsystems. Similar to actual ripples propagating on the surface of water, Ripple\nNetwork stimulates the propagation of user preferences over the set of\nknowledge entities by automatically and iteratively extending a user's\npotential interests along links in the knowledge graph. The multiple "ripples"\nactivated by a user's historically clicked items are thus superposed to form\nthe preference distribution of the user with respect to a candidate item, which\ncould be used for predicting the final clicking probability. Through extensive\nexperiments on real-world datasets, we demonstrate that Ripple Network achieves\nsubstantial gains in a variety of scenarios, including movie, book and news\nrecommendation, over several state-of-the-art baselines.\n

Affiliated Institutions

• Microsoft Research Asia (China) CN
• Hong Kong Polytechnic University HK
• Shanghai Jiao Tong University CN

Related Publications