Improving Memory-Based Collaborative Filtering by Neighbour Selection based on User Preference Overlap

User-based collaborative filtering approaches suggest interesting items to a user relying on similar-minded people referred to as neighbours. While standard approaches select neighbours based on user similarity, others rely on aspects related to user trustworthiness and reliability. We investigate the extent to which user similarities are essential to obtain high quality item recommendation, and pro-pose to select neighbours according to the overlap of their prefer-ences with those of the target user. We empirically show that a neighbour selection strategy based on preference overlap achieves better performance than similarity- and trust-based selection strate-gies, in terms of both recommendation accuracy and precision