Title: Dynamic Clustering Recommendation Algorithm For Two-Layer Graph Attention Network
Abstract: Collaborative filtering recommendation algorithm is one of the most widely used personalized recommendation algorithms in e-commerce websites. The traditional collaborative filtering recommendation algorithm has a high recommendation complexity and low accuracy with the increasing number of users and items in recent years. The previous differential clustering evolution process only recommended a single clustering results of users or items. Besides, the node state of network was only a scalar, which ignored the integration of user layer and item layer and could not better represent the attribute characteristics of users and items. This paper proposes an effective collaborative filtering recommendation algorithm for the above three problems. We fully explore the changes of interests of users and their attention to the items over time. Firstly, a time-weighted scoring matrix is constructed by combining the forgetting function. According to the new scoring matrix, the user-item attention matrix is obtained. Then, according to the differential equations, users and items with high relevance are gathered to obtain the user communities and item communities. Stabilizing the same user status values mean that they have similar interests and then they are assigned to the same community. Finally, the real time prediction results are obtained through improved prediction method and dynamic similarity measurement in each community. The effectiveness of the proposed algorithm is verified by comparison with several better algorithms.
Publication Year: 2019
Publication Date: 2019-10-01
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot