K - Nearest Neighborhood based Music Recommendation System

A. Introduction Recommendation systems are an active topic in research and industry. The technique of collaborative filtering is especially successful in generating personalized recommendations. More than a decade of research has resulted in numerous algorithms, out of which, we have chosen K-Nearest Neighborhood (K-NN) model to predict the ratings for the songs. This model is an item-based algorithm which looks for neighbors among items (songs in this context) unlike user-based algorithms which look for neighbors among users. Our objectives for this project are as follows: 1. To improve prediction results using KNN algorithm. 2. Reduce overall execution time of KNN calculations by introducing parallelization 3. Find the optimum value of K(where K is the number of most similar neighbors chosen to predict the rating for a song for a given user) 4. Improve performance further by using multi cores.