Multi-preference Social Recommendation of Users Based on Graph Neural Network

Abstract

In recent years, social-based recommendation uses social information to alleviate the problems of cold start and data sparseness in traditional recommendation, which is widely used in e-commerce and movie recommendation. However, most of the existing researches make recommendations based on users' short-term preferences or historical records, and fail to fully dig out the user's preference characteristics. This paper proposes a user long-term and short-term preference model based on graph neural network, which uses the nodes of the user's social graph and the user's product graph to aggregate neighbor information, and iteratively updates the feature representation of the target node. The gated recurrent units is used to extract the long-term and short-term preferences of users, and the attention mechanism is used for weight distribution. In addition, we use specific vectors to represent the long-term characteristics of slow changes in users. By conducting experiments on two commonly used data sets, it can be shown that our proposed model is better than the compared baseline method.