A social recommendation method based on social influence propagation

Abstract

The invention discloses a social recommendation method based on social influence propagation, the steps of which include: 1. Constructing a user's rating matrix for items and a social relationship matrix between users; 2. Constructing an initial feature matrix of users and items; 3. According to the K evolution, the contribution of the user's social influence propagation to the user fusion feature matrix is ​​obtained; 4. The contribution of the item to the user fusion feature matrix is ​​calculated according to the items rated by the user's history; 5. The user's contribution to the item is obtained through the matrix inner product operation Predicted scoring matrix. The present invention can alleviate the data sparsity problem in the traditional recommendation model based on social influence, and calculate the contribution of social influence propagation according to multiple evolution operations at the same time, realize the accurate modeling of user fusion feature matrix, and thus realize accurate item selection for users recommend.

Description

technical field [0001] The invention relates to the field of personalized recommendation, in particular to a social recommendation method based on social influence propagation. Background technique [0002] With the rapid development of the Internet, e-retail, community and other websites emerged as the times require. Faced with a huge number of resources, users are often distressed because they cannot find the items they like. Therefore, the recommendation system for users' hobbies has become an essential technology for major websites, which independently recommend items that may be of interest to each user. [0003] Collaborative filtering algorithms are widely used in recommendation systems and have achieved good results. Collaborative filtering algorithms are based on historical interaction data between users and items. Item types can include physical goods, as well as virtual goods such as movies and e-books. The interaction data between a user and an item may include...

AI Technical Summary

Problems solved by technology

Some existing social recommendation works solve the recommendation problem based on collaborative filtering technology, and only consider the influence of the user's social neighbors

However, the spread of users' social influence will also contribute to the modeling of users' interests and hobbies. Therefore, the problem of social recommendation is how to design a model structure to capture the influence of social influence spread on the accurate modeling of their interests and hobbies. Alleviate the data set sparsity problem in traditional recommendation models

Images