Research And Implementation On Social Recommendation Algorithm By Integrating Matrix Factorization With Graph Embedding

With the rapid development of information and internet technology,human society have entered the era of information overload.In this context,it is difficult for users to find the information they are interested in,and the information itself is difficult to show to users who may be interested in it.The recommendation system is a technology proposed to solve the information overload.Due to the problems of data sparsity and cold start in traditional recommendation systems,social information is introduced.From the perspective of heterogeneity,it reflects the indirect relationship between users,and from the perspective of homogeneity,it reflects the direct relationship between users.At present,most of the recommendation systems based on social information are considered from the perspective of heterogeneity or homogeneous,so how to extensively exploit and comprehensively explore the homogeneous and heterogeneous information in social network for recommendation becomes a challenge.In addition,the existing recommendation algorithms rarely consider the influence of time factors on the recommendation results.To solve these problems,this thesis combines graph embedding technology and matrix factorization to conduct an in-depth study on social recommendation system.The main work is as follows:1.To fully explore the deep structure hidden by information network,we propose a unified H2 Rec model to fuse homogeneous and heterogeneous information for recommendations in social networks.Considering the rich semantics reflected by metapaths in heterogeneous information and the wealth of social information reflected by homogeneous information,the proposed method uses a random walk strategy to generate node sequences in a homogeneous information network and a random walk strategy guided by metapaths to generate node sequences in a heterogeneous information network.Finally,we combine the two parts into a unified model for social recommendation.2.Considering the influence of time factor on the recommendation results,we first define the time-based co-occurrence network,and then construct a users co-occurrence network and items co-occurrence network through user-item interaction data.The user co-occurrence network,item co-occurrence network and social network are pre-trained to get embedded representation of nodes and exploit the deep structure in information network.Finally,a recommendation model Co NERec based on co-occurrence network embedding is proposed,which combines embedding information with matrix factorization for recommendation.3.Design and implement a music recommendation system that can provide users with personalized music recommendation services based on the algorithms proposed in this paper.