An efficient system using item & user-based CF techniques to improve recommendation

Nowadays large portion of web-based businesses, research projects, and scientist use recommendation systems to help their business to thrive & flourish. Standard recommendation system utilises either user CF, item CF or content based recommendation system, these furthermore confront issues like item cold start, user cold start and real-time prediction problem. In the perspective of these challenges, cluster based hybrid CF approach is proposed in this paper which uses item-based CF algorithm combined with user demographic based CF algorithm in clusters weighted mechanism. The proposed system is adaptable and extendable which is fruitful in addressing not only user cold start issues but also item cold start issues along with sparsity problem, with lower MAE enhancing the structure to give better suggestion progressively.