Movie recommendation method combining knowledge graph with auto-encoder

Abstract

The invention discloses a movie recommendation method combining a knowledge graph with an auto-encoder. The movie recommendation method comprises the following steps of: 1, obtaining a movie languagecategory of a movie from a knowledge graph DBpedia as additional information; 2, performing vectorization representation on the obtained movie language category information by using a multi-hot methodand taking the information as initial feature extension; 3, performing dimensionality reduction on the obtained initial feature extension by using an auto-encoder, and taking the encoding layer output of the auto-encoder as the extracted low-dimensional feature representation; and 4, fusing the obtained low-dimensional feature representation into an original feature space of the movie, and inputting a new feature as additional information into the semi-auto-encoder model to realize more accurate movie recommendation. According to the method, the knowledge graph can be used for carrying out feature extension on the movie information, the extended features are processed through the self-encoding machine, high-level low-dimensional feature representation is obtained so as to be input into the recommendation model for prediction, and the purpose of carrying out more accurate recommendation for the user is achieved.

Description

technical field [0001] The invention relates to the field of personalized data recommendation research, in particular to a movie recommendation method using a knowledge map combined with an autoencoder. Background technique [0002] Nowadays, with the explosive growth of data information, people are faced with a large number of service and commodity choices, which also makes it more difficult for users to efficiently find useful information and products for them. Recommender systems can help people alleviate the problem of information overload. Among various recommendation methods, collaborative filtering recommendation algorithm has achieved remarkable results in recent decades. However, it faces the problem of sparse scoring matrix and weak generalization ability. In order to solve these problems, matrix factorization technology was proposed, the main operation is to learn the hidden features of users or products from the rating matrix to optimize the recommendation accu...

AI Technical Summary

Problems solved by technology

Although these methods can use additional information of users or products, it is difficult to obtain auxiliary information from other information sources, and most of the auxiliary information itself is very sparse, so it is difficult to effectively and directly integrate into the recommendation system to improve the recommendation accuracy.

Images