Basic Consideration of Rough C-Means Clustering-Based Ensemble Collaborative Filtering

Abstract

In clustering-based collaborative filtering (CF), clusters composed of users with similar preference patterns are extracted, and items with high preference within the cluster are recommended. Since data in CF tasks contain uncertainties due to human sensibility, rough clustering based on rough set theory, which deals with uncertainties, is considered effective. In this context, RCM-CF, a CF approach based on rough C-means (RCM), a type of rough clustering, has been proposed. In this study, we propose RCM-based ensemble CF (RCM-ECF), which is an improved version of RCM-CF incorporating ensemble learning. RCM-ECF makes final recommendations by integrating multiple recommendation degrees derived from multiple clustering results due to different hyperparameter settings. Furthermore, we verify the recommendation performance of the proposed method through numerical experiments using real-world datasets.