A novel approach for improving recommender system

Abstract

Recommender systems these days tend to achieve status and importance. The algorithms and approaches present in a recommender system define the importance and worth of the recommender systems. Till now, the research which have been done has concentrated on generating the new approaches and algorithms so that by implementing them some good recommendations can be provided to the users. In this research paper we propose that a recommender system can be developed efficiently so that it can work beyond the conventional accuracy conditions and this can be done by comparative study of any two algorithms and generating a new algorithm based on the results of both algorithms which would give some more accurate results. Experimental results specifies that the people with different views mostly tends to favor different types of algorithms. There are few criteria for evaluating the recommender systems which can be helpful in judging the performance.