Graph neural network federal recommendation method for privacy protection

Abstract

The invention discloses a graph neural network federal recommendation method for privacy protection. At present, many academic strategies provide recommendation methods such as matrix decomposition, collaborative filtering and the like to improve the recommendation accuracy, but in the proposed methods, the recommendation accuracy, the problem of data islands, the problem of how to jointly train a plurality of clients, and the security and privacy of data cannot be taken into account at the same time. The method comprises three parts of contents: performing graph neural network recommendation based on each client of a bipartite graph, performing joint training of a graph neural network recommendation method based on federated learning, and performing homomorphic encryption of server and client transmission data facing privacy protection. According to the graph neural network federal recommendation method for privacy protection, a plurality of clients can be combined in a data island environment to carry out graph neural recommendation modeling training for ensuring privacy and data security, so that the recommendation accuracy of all the clients is remarkably improved, and the data security is protected.

Description

technical field [0001] The invention belongs to the field of federated recommendation systems, and relates to a privacy protection-oriented graph neural network federated recommendation method, especially a method requiring high data security protection. Background technique [0002] In recent years, with the rapid development of technologies such as cloud computing, big data, and the Internet of Things, the emergence of various applications in the Internet space has triggered an explosive growth in data scale, and more and more information and services are flooding the Internet. More and more information can be exposed to in daily life, but at the same time it also increases the difficulty of finding useful information for oneself, that is, "information overload". The user's knowledge level and cognitive ability are limited. When faced with massive and complex Internet information, they cannot quickly find the information they need, or even understand and use the informatio...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem of how to combine multi-party data to give full play to the advantages of big data and ensure data security for recommendation under the dilemma of data islands, and propose a privacy-oriented graph neural network federated recommendation (FGC) method

Images