Review on “Typicality-Based Collaborative Filtering Recommendation using Sub Clustering for Online Shopping”

Abstract

Collaborative filtering is a convenient mechanism used in recommender system, which is used to find the similar items in a group. The same favour items can be identified by using the collaborative filtering based on items and the users. However there are some drawbacks in premature filtering techniques which lead to less accuracy, data sparsity and prediction errors. In this work take advantage of proposal of object typicality from cognitive psychology moreover suggests a typicality-based collaborative filtering recommendation method named as Tyco. A distinguishing characteristic of typicality-based collaborative filtering is that it finds neighbours of users on the basis of user typicality degrees in user groups. Selection of neighbours regarding users by means of measuring users’ similarity on the basis of their typicality degrees is a separate feature, which distinguishes this approach from earlier collaborative filtering methods. It exceeds many CF recommendation methods on recommendation accuracy on any type of datasets. In proposed method main approach is to Sub Clusters the all items into several item groups by applying such as nearest neighboring algorithm. This helps users to search items more easily and to increase the accuracy and quality of the recommendation.