Scalability Improvement using Map Reduce Algorithm in Recommender Systems

Abstract

The proficiently-liked technology for recommender system is collaborative filtering. The current CF methods struggle problems with recommendation inaccuracy, data sparsity and errors in prediction. For curt data retrieval, the implementation of cluster along with map reduce algorithm can glide exactness in prediction and scalability. Clustering of all items into a group is made and then the formation of user group corresponding to each item group is done. By now all the users having swing typically degrees in each of the user group is made. The user typicality matrix to sham the adherent similarities is built. This fanatic typicality matrix based approach will lead to pick a set of neighbours of each user. The prediction of everyday rating of a user concerning the order of an item based upon the ratings of neighbours at adherent upon the item.