Film individuation recommendation method based on user real-time interest vectors

Abstract

The invention discloses a film individuation recommendation method for combining film content and user real-time scoring information. The problem that a traditional recommended algorithm cannot reflect user interest change and data sparsity in time is mainly solved. In order to solve the data sparsity problem, the user interest vectors are introduced in the film individuation recommendation method. Starting from the film feature vectors, the obtained user interest feature vectors are processed in an iterative mode by the aid of a user scoring matrix, a user similar matrix is established according to the obtained user feature vectors, and finally recommendation can be achieved according to a traditional collaborative filtering scoring predicator formula. According to the user interest change condition, time factors are further integrated in the establishing process of the user interest vectors, the scoring behavior weight is bigger when scoring time more approaches the current time, and the user interest can be more represented.

Description

technical field [0001] The invention is a recommendation method generated by synthesizing real-time rating data of users and feature attributes of movies, which mainly solves the problems of data sparsity and inability to reflect user interest changes in time in traditional collaborative filtering recommendation algorithms, and belongs to the field of multimedia information processing. Background technique [0002] As the Internet has brought us into the era of information explosion, movie content has also grown rapidly in recent years, and countless websites and applications related to movie content have been launched. For users, in the face of such a wealth of movie resources, it is very difficult to pick out the content they really want. The recommendation system has become an effective way to solve these problems, and the recommendation algorithm is the key to the realization of the function of the recommendation system. The current mainstream recommendation algorithms ...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of data sparsity and inability to reflect changes in user interests in the traditional collaborative filtering recommendation algorithm, the present invention proposes a recommendation method based on real-time user interest vectors that integrates real-time rating data of users and feature attributes of movies

Images