Web based recommendation system using Multi-attribute collaborative filtering for user satisfaction

Abstract

In the wide area of personalized user based recommendation, although resource attribute is one the biggest important factors in recognizing user preferences, the researchers take into consideration among the user interest differences in resource attribute. As per previous research, both prediction and similarity computation are not extremely precise. There are some areas which have some space for improvements. In terms of accuracy, this paper proposed a modified ratio based multi-attribute method to calculate the similarity, providing us a new evaluation model of user interest based on resource multi-attribute. By comparing the multi attribute values we can find similarity between users and items using PARAFAC algorithm which is also responsible to handle large dataset and parallel computation among multi-attribute. This proposed method is also able to evaluate performance using this large data set of real web services whose experimental results explain that proposed method, in this paper, achieve better prediction and take less time in computation than various references considered.