Film Recommendation System with Social-Union Algorithm: Film Recommendation System with Social-Union Algorithm

Abstract

A Recommendation System applies several classic Collaborative Filtering (CF) methods. Some CF methods are combined with social networks, for instance Fusing ESR, Social Regularization, and Trust-Aware. Nonetheless, these three methods are not able to be developed if they are integrated with other types of implicit or explicit relationships. Their selection of parameter weights is not optimal enough when it combines preferences between users from the two types of relationships into one. Usually, a group of friends will be similar in terms of interest and preference for an item. The similarities between users will increase the accuracy of the prediction results. The selection of parameter weights can be done manually or automatically through the calculation of global and local density coefficients so the determination of parameter weights will be optimal. Therefore, the Social-Union (SU) method proposed in this research use that method to overcome the problems from previous research. The result of this research is a website that applies the Social-Union method and let the user get recommendations that depend on the value of the parameter a.