A Knowledge Graph-Enhanced Attention Aggregation Network for Making Recommendations

In recent years, many researchers have devoted time to designing algorithms used introduce external information from knowledge graphs, solve the problems of data sparseness and cold start, thus improve performance recommendation systems. Inspired by these studies, we proposed KANR, a graph-enhanced attention aggregation network for making recommendations. This is an end-to-end deep learning model using graph embedding enhance It consists three main parts. The first network, which collect user’s interaction history captures preference each item. second graph-embedded model, aims integrate knowledge. semantic nodes edges in mapped low-dimensional vector space. final part unit, fusing features two vectors. Experiments showed that our achieved stable improvement compared baseline recommendations movies, books, music.