{"title":"Local Matrix Factorization with Social Network Embedding","authors":"Jinmao Xu, Shuaiheng Peng, Daofu Gong, Fenlin Liu","doi":"10.1109/AIID51893.2021.9456514","DOIUrl":null,"url":null,"abstract":"In the recommender system, how to construct submatrices for local matrix factorization is an important problem. In this paper, we propose the Local Matrix Factorization with Social Network Embedding (LMFE) method in order to construct more meaningful sub-matrices and improve the performance of the recommender system. Firstly we utilize the user's social information and rating information to construct a heterogeneous information network (HIN). And then extract the node representations of users and items from HIN. We use the representations of the node as the basis for sub-matrix division. Finally, the local matrix factorization is performed on sub-matrix to obtain the prediction results. Experimental results from the real-world dataset Yelp demonstrate that the LMFE can achieve better performance than the comparative method.","PeriodicalId":412698,"journal":{"name":"2021 IEEE International Conference on Artificial Intelligence and Industrial Design (AIID)","volume":"362 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE International Conference on Artificial Intelligence and Industrial Design (AIID)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/AIID51893.2021.9456514","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
In the recommender system, how to construct submatrices for local matrix factorization is an important problem. In this paper, we propose the Local Matrix Factorization with Social Network Embedding (LMFE) method in order to construct more meaningful sub-matrices and improve the performance of the recommender system. Firstly we utilize the user's social information and rating information to construct a heterogeneous information network (HIN). And then extract the node representations of users and items from HIN. We use the representations of the node as the basis for sub-matrix division. Finally, the local matrix factorization is performed on sub-matrix to obtain the prediction results. Experimental results from the real-world dataset Yelp demonstrate that the LMFE can achieve better performance than the comparative method.