Improving Stability, Smoothing and Diversifying of Recommender Systems

Sagar Sontakke, Pratibha Chavan

Abstract: Recommender systems are used extensively now-a-days for various web-sites such as for providing products suggestions based on customers purchase history and searched product s. Current recommendation system approaches lack of a high degree of stability. Diversification of prediction is also important feature of recommender system. Having displayed same set of results every time may increase lack of trust in recommendation systems. In this paper, our focus is on stability of different recommender system approaches and providing diversity in results. We will focus on different recommender system approaches, methods provided to improve the stability and approaches in increasing diversity of recommender system.

Keywords: Recommendation Systems, Stability, Iterative Smoothing, Collaborative Filtering, Jaccard Distance, Clustering