A Recommended Method Based on Semantics and User Preferences in Network Live Television

This paper proposes a real-time recommendation method based on semantics and user preference in Live TV. Firstly, it constructs a user preference model based on users' watching history, and calculates the similarity between the to-be-recommended programs with users' watching history or users' watching programs respectively with the Semantic similarity calculation method based on word vector. Combined with user preference, the similarity is used to work out users' virtual interest about the to-be-recommended programs. Finally, it's the program which gets higher virtual interest that becomes users' real-time recommendation result. Experimental results show that the recall rate of this method is improved by about ten percent than the existing methods when it's used on the real-time predicted recommendation. And the quality of recommendation is improved significantly on real-time program recommendation.