A Personalized Recommender Algorithm Based on Fuzzy Relation Reputation Model

Abstract

For overcoming the problems that the traditional collaborative filtering personalized recommender algorithm which called KNN has such as cold-starting, data sparsity, flexibility and black box, a new personalized recommender algorithm based on fuzzy reputation model(called FRPRA) is proposed. We analyze the steps of FRPRA and the differences between it and KNN, explore the ways how FRPRA overcomes the existed problems of KNN. At the same time, we compare the performance of this two algorithms.