COLLABORATIVE FILTERING RECOMMENDATION ALGORITHMS TO ALLEVIATE SPARSITY AND SCALABILITY PROBLEMS

Abstract

The Internet and World Wide Web have given us a world of endless possibilitieslike items to consume, movies to watch, music to listen, conversations to participate in etc. Amidst all this range of endless options, a consumer faces the task of what to choose which might interest him. Recommender system comes to the rescue for such a consumer. These systems aim to mediate, support, or automate the everyday process of sharing recommendations. Recommender systems are making their presence felt in a number of domains, be it for ecommerce or education, social networking etc. With huge growth in number of consumers and items in recent years ,recommender systems faces some key challenges. These are: producing high quality recommendations and performing many recommendations per second for millions of consumers and items. New recommender system technologies are needed to reduce sparseness in order to get high quality recommendations, even for very large-scale problems. In this thesis, we focus on collaborative approach based recommender systems to solve the issues of sparsity and scalability. We have compared two collaborative filtering algorithms which solves the above mentioned issues in one go. The first algorithm uses weighted slope one method to reduce sparseness and the other one uses item classification technique. Then item clustering is used to alleviate the scalability problem. The item classification technique is better among the two as determined using the case study from movielens data set obtained from Grouplens website.