Program data recommendation method and system

Abstract

The invention provides a program data recommendation method and system, and the method achieves the recommendation of program data based on a knowledge graph and graph learning enhanced representation, and comprises the steps: extracting training data and verification data from a preset user program viewing database; training a preset NPR-KGER model according to the training data, and determining a combination mode of parameters and information of the NPR-KGER model according to the verification data and the trained NPR-KGER model; determining mold entering parameters of the NPR-KGER model according to the combination mode of the parameters and the information; and predicting and recommending programs watched by the user based on the in-mold parameters and the NPR-KGER model. According to the method, the NPR-KGER algorithm is provided, 2-hop set representation of neighbor users and neighbor programs is learned through the knowledge graph, program information and a neighbor program set as well as historical watching behaviors of the users and a neighbor user set are fused, enhanced representation of the programs and the users is achieved, the dynamism and generalization of the model are comprehensively improved, and the user experience is improved. And the accuracy and the sorting index value of the recommendation system are improved.

Description

technical field [0001] The present invention relates to the field of intelligent recommendation in the field of artificial intelligence, and more particularly, to a method and system for recommending program data based on knowledge graph and graph learning enhanced representation. Background technique [0002] As the Internet gradually enters the stock era, the importance of recommendation is further increasing. After the flood of data in the information world, the recommendation algorithm can more efficiently acquire value increments for a relatively fixed user group. The recommendation algorithm has been tested by the market and has been widely used in many fields such as music recommendation, e-commerce, medical recommendation, news, and movies. It has brought huge convenience and benefits to many Internet companies and even other organizations, and has gradually become a An integral part of the Internet. [0003] There are various existing program recommendation algorit...

AI Technical Summary

Problems solved by technology

However, the existing radio and TV program recommendation is not deep enough in using auxiliary information to alleviate data sparsity, and it is not comprehensive enough to learn users' dynamic interests and generalized interests, and there is still room for improvement and optimization

Images