Group-based collaborative filtering supported by multiple users' feedback to improve personalized ranking

Recommender systems were created to represent user preferences for the purpose of suggesting items to purchase or examine. However, there are several optimizations to be made in these systems mainly with respect to modeling the user profile and remove the noise information. This paper proposes a collaborative filtering approach based on preferences of groups of users to improve the accuracy of recommendation, where the distance among users is computed using multiple types of users' feedback. The advantage of this approach is that relevant items will be suggested based only on the subjects of interest of each group of users. Using this technique, we use a state-of-art collaborative filtering algorithm to generate a personalized ranking of items according to the preferences of an individual within each cluster. The experimental results show that the proposed technique has a higher precision than the traditional models without clustering.