Method and system for movie recommendation based on visual features

Abstract

The invention discloses a movie recommendation method based on visual features, which includes: according to the feature factor of the user to be recommended, the feature factor of the unrated movie, the similarity coefficient between the unrated movie and other movies, pre- The extracted visual features of the unrated movie and the weights of each feature in the visual features are used to predict the predicted score of the unrated movie by the user to be recommended by using a pre-established movie prediction model; wherein, the Visual features include color histogram, SIFT feature, CNN feature and movie category feature; judge whether to recommend the unrated movie to the user to be recommended according to the predicted score. In addition, the embodiment of the present invention also discloses a movie recommendation system based on visual features. By adopting the embodiments of the present invention, the accuracy of movie recommendation can be improved, and user experience can be improved.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and system for recommending movies based on visual features. Background technique [0002] Movie recommender systems are one of the popular forms of recommender systems today. The working principle of movie recommendation is to predict the movies that the user may like by analyzing the user's historical movie viewing behavior. However, due to too little historical behavior data of users, the recommendation effect is poor. This phenomenon is called the data sparse problem. Among today's recommendation algorithms, matrix factorization is one of the most effective methods proven so far. According to the difference of the used data, the matrix factorization technology in the recommendation algorithm is divided into matrix factorization without context and matrix factorization based on context. [0003] A matrix factorization method without context can be describ...

AI Technical Summary

Problems solved by technology

In fact, this assumption only imposes certain restrictions on the learning process of the factor matrix

Therefore, although the existing context-based matrix factorization method can improve the recommendation effect and performance to a certain extent, it cannot guarantee the improvement of recommendation accuracy.

Images