Personalized Microblog Recommendation System Integrates Following and Reposting Relationship

At present, the research on personalized recommendation methods mainly uses user tags and friends to explore user interests, and does not effectively utilize the user's active interaction information such as following relationship and reposting relationship. Additionally, data will be more sparse after increasing relationships, only using methods based solely on binary relations, such as collaborative filtering, cannot achieve the ideal recommendation effect. To deal with such problems, this paper proposes a microblog personalized recommendation system based on lightweight graph convolutional neural network that integrates attention relationship and forwarding relationship. First, the user's following relationship and reposting relationship are integrated into the user and microblog embedding to enhance the user representation vector and the microblog representation vector, and then the microblogs will be accurately recommended to the user through recommendation score ranking. The experimental results show that the performance of the model has been significantly improved after integrating following information of users and reposting relationship of microblogs, and the cold start problem can be alleviated at the same time.