Recommender Systems: An Algorithm To Predict "who Rate What"

Recommender systems are systems that recommend content for us by looking at certain factors including what other people are doing as well as what we are doing. Examples of such systems present today are Amazon.com recommending books, CDs, and other products; Netflix recommending movies etc. These systems basically recommend items or movies to customers based on the interests of the present customer and other similar customers who purchased or liked the same item or movie. Our paper goes beyond the concept of overall generic ranking and provides personalized recommendation to users. Despite all the advancements, recommender systems still face problems regarding sparseness of the known ratings within the input matrix. The ratings are given in the range of (1-5) and present systems predict What are the ratings'' but here we propose a new algorithm to predict Who rate what'' by finding contrast points in user-item input matrix. Contrast points are the points which are farthest from the known rated items and most unlikely to be rated in future. We experimentally validate that our algorithm is better than traditional Singular Value Decomposition (SVD) method in terms of effectiveness measured through precision/recall.