An Attention Mechanism Fusion Knowledge Graph Movie Recommendation Method

Abstract

The present invention relates to the technical field of recommendation systems, in particular to a movie recommendation method that integrates the attention mechanism of the knowledge map, including: acquiring user-movie interaction information and movie attribute information, and performing data preprocessing through a DAE noise reduction automatic encoder; The processed data set constructs a knowledge graph; models the knowledge graph to obtain entity vectors and relationship vectors; constructs a deep learning hybrid recommendation system with a memory component attention mechanism to obtain user vectors and item vectors; finally, through joint learning, Combining the knowledge map with the hybrid recommendation system, the user's attention score to the candidate movies is obtained through the attention mechanism, and the predicted score value is obtained by sorting in descending order, and then TOP-N recommendation is made.

Description

technical field [0001] The invention relates to the technical field of recommendation systems, in particular to a movie recommendation method that integrates an attention mechanism of a knowledge graph. Background technique [0002] With the rapid development of the mobile Internet and the wide application of e-commerce, various entertainment website platforms have emerged one after another, such as movie websites, music websites, Douyin and other short videos. However, with the development of the information age, most websites are facing the problems of information overload and information lost, so to address these two problems, the recommendation system came into being. [0003] Among them, the recommendation system also plays an important role on various Internet platforms. Through the recommendation system, users can more efficiently and quickly find products or information that they are interested in or have potential interests in. Based on the user's historical behavi...

AI Technical Summary

Problems solved by technology

[0006] At the same time, in the field of video websites, the phenomenon of low recommendation accuracy and insufficient explainability generally exists. Therefore, this invention proposes a hybrid recommendation system using knowledge graphs and attention mechanisms for movie websites. The knowledge added Atlas can effectively improve recommendation accuracy and have a certain interpretability; the attention mechanism with memory components in deep learning can better capture the dynamic changes of user interests and improve operating efficiency, thereby appropriately improving recommendation accuracy.

Images