Recommendation method based on graph twin network

Abstract

A recommendation method based on the graph twin network is applied to the field of personalized recommendation. An existing method (1) lacks knowledge expansibility, for example, social relation information of users is difficult to fuse effectively; (2) under the propagation normal form of the multilayer feature information, the learned features have an over-smooth problem; therefore, the invention provides the recommendation method based on the graph twinning network, the user relationship graph and the article relationship graph are modeled through the interaction information of the user andthe article, and the user relationship information and the article relationship information are respectively mined in the form of two isomorphic directed graphs through the graph convolution layer designed by the invention. And finally, the user features and the article features of the two channels are aggregated through a graph interaction layer, and user preference information and article attribute information are fully extracted. According to the method, the UI feature characteristics are effectively maintained, the personalized recommendation accuracy is remarkably improved, and the method has good model expandability and wide application prospects.

Description

technical field [0001] The invention is applied to the field of recommendation system based on U-I relationship, and specifically relates to data mining and deep learning technologies such as graph convolutional neural network, attention mechanism, user preference information and item attribute information feature extraction, U-I interactive information modeling, etc. Background technique [0002] Personalized recommendation is a comprehensive analysis task, widely used in social network, music radio station, e-commerce, personalized advertisement, movie and video website, etc., so it has attracted much attention. In recent years, deep learning has achieved great success in many research fields such as computer vision and natural language processing, and has aroused great interest. Deep learning is widely used in recommendation systems, and its effectiveness has been proved in the fields of information retrieval and recommendation system research, and it is the development d...

AI Technical Summary

Problems solved by technology

However, most of the above methods model user item information into a graph, which makes it difficult to effectively control the complexity of the model and the model training time. Moreover, the recommendation method based on a single graph lacks knowledge scalability, and it is difficult to aggregate multi-platform U-I relationships. Other information sources such as user social relations

Images