Matrix decomposition and collaborative filter algorithm combined movie recommendation method

Abstract

The invention discloses a matrix decomposition and collaborative filter algorithm combined movie recommendation method. The method comprises the following steps of: obtaining a user original scoring matrix, wherein the user original scoring matrix comprises scoring information, for M movies, of N users; extracting a user-feature matrix UZ on the basis of the user original scoring matrix by adoption of singular value decomposition; calculating a user similarity matrix SIMu, v on the basis of the user-feature matrix UZ; calculating k nearest neighbor feature set of the users on the basis of a preset k value and the user similarity matrix SIMu, v; calculating a predicted score of each movie on the basis of the k nearest neighbor set of the users and the user original scoring matrix; and sorting and recommending all the movies according to a preset law on the basis of the predicted score of each movie. According to the method, the recommendation correctness of a slope one algorithm in movie recommendation systems can be enhanced, the personalization degree can be improved while the operation speed of the algorithm is ensured, and influences caused to the slope one algorithm by matrix sparseness can be decreased.

Description

technical field [0001] The invention belongs to the technical field of movie recommendation, and in particular relates to a movie recommendation method combined with matrix decomposition and collaborative filtering algorithm. Background technique [0002] Using collaborative filtering algorithm for recommendation is the most mature and general method in the field of recommendation system. Traditional collaborative filtering algorithm is divided into item-based collaborative filtering (Item-based Collaborative Filtering) and user-based collaborative filtering (User-based Collaborative Filtering). ). Slope One is an Item-Based collaborative filtering recommendation algorithm. The main idea of ​​the algorithm is to use the rating deviation of all users to predict the rating of a specific user. The idea of ​​this algorithm is easy to understand, and can be easily implemented on various platforms. It is superior to traditional collaborative filtering algorithms in terms of accur...

AI Technical Summary

Problems solved by technology

In order to solve this problem, one method is to perform cluster analysis on users before prediction, but considering that the user-movie matrix is ​​usually large in size, the time and space complexity required for clustering in large-scale data sets is too high, using This method is not suitable for fast recommendation. One method is to use label-based slope one recommendation. This method makes up for the lack of personalization to a certain extent, but because users rate movies is a subjective process, usually these Tags do not fully represent key features of users or movies

Images