An Improved Collaborative Filtering Based Recommender System using Bat Algorithm

Abstract

Recommender Systems have proven to be of great aid in dealing with the issue of Information Overload by improving the user experience through quality recommendations. In recent times, heuristic techniques have been employed by researchers in recommender systems along with traditional methods of collaborative and content based filtering. On the same account, in this work a Bat algorithm based heuristic technique has been used to compute the weights of items (features) so as to find better neighbourhood for the active user. We argue and also prove using the results that this technique of giving weights to items using heuristic methods helps in achieving better personalized recommendations. The performance of this system was also compared to that of Artificial Bee Colony based system (ABC). The results indicated that BA performed 6.9% better than ABC in terms of Mean Absolute Error and F1 Score using our technique.