Collaborative filtering recommendation method and device based on knowledge graph and deep learning

Abstract

The invention discloses a collaborative filtering recommendation method and device based on a knowledge graph and deep learning, and the method comprises the steps: obtaining the relation data of a user and a project, and building a user-project interaction graph and a knowledge graph; generating an embedded matrix according to the user-project interaction graph to obtain original embedded vectors of the user and the project, and extracting a triple in which entities associated with the project in the knowledge graph are located to obtain embedded vectors corresponding to the entities in the triple; recursively propagating and embedding the user embedded vectors in an L-layer Light-GCN network to obtain user embedded vectors of each layer, and combining the user embedded vectors to obtain user high-order embedded vectors; spreading and embedding the knowledge graph by means of ripple networks, and obtaining a project high-order embedded vector through propagation of L ripple networks and high-order interaction of a cross compression unit; and transmitting the user high-order embedded vector and the project high-order embedded vector into a dot product prediction function, and recommending an interested project to the user according to a prediction result. The method overcomes the defects of cold start and sparsity of a traditional recommendation method.

Description

technical field [0001] The invention relates to the field of recommendation, in particular to a collaborative filtering item recommendation method and equipment. Background technique [0002] With the rapid development of Internet technology, the Internet is more and more involved in people's work and life. For example, you can buy goods through online stores, participate in various activities through the Internet, and search for interesting information such as movies, books, and knowledge points. Wait. As the amount of information increases, many recommendation systems have emerged to recommend items / items that fit users' interests. Recommender systems generally use recommendation algorithms to obtain data and predict which options a given group of users will be interested in. Collaborative filtering is an algorithm that addresses personalized recommendations by assuming that users with similar behaviors will show similar preferences for items. However, in the Internet a...

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: Aiming at the problems of the prior art, the present invention proposes a collaborative filtering recommendation method based on knowledge graph and deep learning, which overcomes the defects of poor interpretability and low accuracy of traditional recommendation methods

Images