Enhancing Prediction in Collaborative Filtering-Based Recommender Systems
Majid Hatami, Saeid Pashazadeh · 2014
Recommender systems (RS) are introduced to help users with finding the desired information. Collaborative filtering (CF) approach is one of the most widely used techniques in recommender systems. Prediction is the main part of all recommender systems. An enhanced prediction formula that could be employed in all CF-based methods is proposed in this paper. Resnick prediction formula that is the most well-known and employed formula in CF-based RS is used as basis in this paper. Not only the average of active user's ratings, but also the collective average of similar users' ratings and the average of all ratings given to the target item is used in the formula of this study. The results are promising and satisfactory. Results of enhanced prediction formula are compared with the results of unenhanced version to verify the effectiveness of proposed method.