Recommendation Systems: Past, Present and Future

Abstract

Every customer want to buy his product having preferred by all his friends in surrounding environment. User communicates to the surrounding people regarding all purchases and give extreme importance to these people's choice, views and preferences. In today's world of competitive environment, surplus amount of products information is available in terms of ratings and reviews on all shopping sites. Before purchasing product, People often like to go through product reviews mentioned on websites. This data of reviews has increased terrifically and it is not easy to collect, store and analyse these reviews within a “tolerable elapsed time”. Therefore, optimal recommendation system is required which will analyse product data based on ratings and reviews. Collaborative filtering will make use of user-item rating matrix given by the user to calculate user and item similarity. Alongwith the analysis of clustered reviews of user's neighbours, these rating similarities will help to give optimized recommendation. Thus it will give strong confirmation to avoid irrelevant recommendation. Also it will provide strong solution to cold start problem.