An Enhanced Recommendation System using proposed Efficient K Means User-based Clustering Algorithm

With the blossoming of electronic commerce, recommendation system helps people to find captivating things easily. Plenty of recommendation systems use collaborative filtering technique which has been evidenced to be one of the most efficacious technique in recommendation systems. But the existing technique suffers from the efficiency issue, as it has to determine the similarity index between each pair of users to find out their neighborhood and scalability issue, where many recommendations has to be executed per second for millions of users and products. Also, the another major problem in existing user based recommendation system is to calculate distances which includes large amount of mathematical operations for finding the best neighbors. This paper presents an implementation of proposed efficient K-means clustering algorithm which shows experimentally the comparison with the existing collaborative filtering technique and existing k means technique in recommendation system in terms of computational time, iterations and root mean square error. The Experimental results shows that proposed recommendation system using efficient k means clustering is more effective, scalable and accurate.