Knowledge graph driven personalized accurate recommendation method

Abstract

The invention provides a knowledge graph driven personalized accurate recommendation method. The method comprises the steps of obtaining related knowledge of an article from a knowledge base accordingto historical behaviors of users, constructing a knowledge graph, initializing vector representation of each node and connection, and determining a feeling domain of each node; generating a trainingsample according to the historical behaviors of the users, and initializing vector representations of all the users and articles; obtaining the feeling domain of the corresponding entity of the articles in the training sample in the knowledge graph, and taking the feeling domains and the sample as graph neural network model input to obtain a possibility prediction value of interaction between theusers and the articles; optimizing model parameters by minimizing a loss function; and after the model optimization process is finished, sorting the prediction values of the possibility of interactionbetween a certain user and all the articles to obtain the recommendation list of the user. According to the method, the knowledge graph information is utilized, the sparsity of historical behavior information of an original user is made up, the users and the articles are described from the multi-dimensional perspective, and the personalized recommendation result is more accurate.

Description

technical field [0001] The present invention relates to the field of machine learning, in particular to a personalized and accurate recommendation method driven by knowledge graphs. Background technique [0002] With the development of information technology and the Internet, people have gradually entered the era of information overload from the era of information scarcity. In this era, both information consumers and information producers are facing a huge challenge: as an information consumer, how to find interesting information from a large amount of information is very difficult. As information producers, how to make the information they produce stand out and attract the attention of users is also a very difficult thing. Recommender systems are an important tool to resolve this contradiction. As an information filtering system, the recommendation system recommends to users the information that they are most likely to be interested in through the user's historical behavi...

AI Technical Summary

Problems solved by technology

Although these models have been proven to be effective on some public datasets, the interaction of nodes in the model is not sufficient, making the node representation in the knowledge graph inaccurate

Images