News Recommendation System Based on Topic Embedding and Knowledge Embedding

News recommendation system is designed to deal with massive news and provide personalized recommendations for users. Accurately capturing user preferences modeling users the key recommendation. In this paper, we propose a new framework, based on topic embedding knowledge (NRTK). NRTK handle titles that have clicked from two perspectives obtain representation :1) extracting explicit latent features mining users' them in historical behaviors; 2) entities propagating potential graph. Experiments real-world dataset validate effectiveness efficiency of our approach.