An Evidential Clustering for Collaborative Filtering Based on Users' Preferences

Users are often surrounded by a large variety of items. For this purpose, Recommender Systems (RSs) have emerged aiming to help and to guide users towards items of interest. Collaborative Filtering (CF) is among the most popular recommendation approaches, which seeks to pick out the most similar users to the active one in order to provide recommendations. In CF, clustering techniques can be used for grouping the most similar users into some clusters. Nonetheless, the impact of uncertainty involved throughout the clusters' assignments as well as the final predictions should also be considered. Therefore, in this paper, we propose a clustering approach for user-based CF based on the belief function theory. This theory, also referred to as evidence theory, is known for its strength and flexibility when dealing with uncertainty. In our approach, an evidential clustering process is performed to cluster users based on their preferences and predictions are then generated accordingly.