Parallel Ratio Based CF for Recommendation System

Abstract

With the increase in E-commerce, Recommendation Systems are getting popular to provide recommendations of various items (movies, books, music) to users. To build the Recommendation System (RS), Collaborative Filtering (CF) techniques are proven efficient. From the main two Collaborative Filtering techniques i.e. User-Based and Item-Based, survey suggest that Item-Based CF provides better recommendations. A novel approach, Ratio-Based CF provides recommendation depending upon the item's ratio is more accurate comparatively but face scalability problem. To overcome this problem a parallel approach can be used instead of sequential. Our experiments shows that Ratio Based CF techniques have more accuracy comparatively as well as Parallel (Hadoop) implementation of Ratio Based CF Techniques have drastically reduce the training time (i.e. ratio calculating time between each pair of items) from 90 minutes in Java to 5 minutes in Hadoop for sub-data of MovieLens 100K dataset.