SorrRS: Social recommendation incorporating rating similarity and user relationships analysis
Taiheng Liu, Zhaoshui He, Peitao Wang · 2020 7th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS) · 2020
Recently, collaborative filtering (CF) algorithms have played an important role in recommendation systems. However, the traditional CF algorithms cannot make full use of information (e.g., users' preference, social relationship, and so on) implied in check-ins of users in the recommendation process. In order to address this issue, in this paper, a recommendation framework is proposed by exploiting historical user behavior, the check-in information and user social relationships to improve the precision of recommendation, as follows. First, user reputation can be obtained by iteratively calculating the correlation of historical ratings of the user and intrinsic qualities of items. Second, we employ user reputation and rating similarity in the basic social recommender model. Finally, we use the social connections among users to obtain the social influence of their friends. To evaluate the performance of our method, a series of experiments conducted on Epinions and Ciao data sets show that our approach performs better than several state-of-the-art methods on metrics of RMSE, precision rate, recall rate and F-Measure.