Enhanced Content-Based Filtering Using Diverse Collaborative Prediction for Movie Recommendation

In re-commender system, collaborative filtering or content-based filtering is one of the most popular methods used to predict items of interest for a user. Each method has their own advantage, though individually they possess several limitations. In order to minimize the limitation, we developed a hybrid re-commender system incorporating components from both methods. Our approach includes a diverse-item selection algorithm that uses a diversity metric to select the dissimilar items among the recommended items from collaborative filtering, which together with the input is fed into content-based filtering. We present experimental result on movielens dataset that show how our approach performs better than content-based filtering and Naive hybrid approach.