Recommender systems are powerful personalization tools which have seen widespread adoption across the Internet. However, it is thought that by emphasizing personalization through the optimization of accuracy-driven metrics, the issue of overpersonalization emerges with negative effects on the user experience. An acknowledged effect of this problem is the filter bubble, manifested when the recommendations and consumption cover only a selected portion of the catalogue, causing the user experience to narrow down on the long term. An increasingly popular countermeasure to the problem is offered by diversifying the recommendations even at the cost of reducing the accuracy of the recommender system.
In this thesis the possibilities of developing a solution to address the problem are investigated through the proposal of a recommendation system which implements the diversity by design for a movie application domain.
Building on past research, a user-centric framework to enhance the diversity on four related dimension, namely global coverage, local coverage, novelty and redundancy is presented. The proposed solution is designed to diversify users profiles, modeled on categorical preferences, within the same group in the recommendation filtering.
A proof of concept is developed to evaluate the diversification levels reached through the extraction of diverse within-group users against random extraction, with different levels of user diversity.