A Novel Recommender System Based on Fuzzy Set and Rough Set Theory

Recommender System is an effective means of handling information overload and can provide personalized service as a useful information tool in e-commerce. In this paper, a novel automatic recommender system is proposed based on fuzzy c-means algorithm and rough set theory, including three main steps: data discretization, rules establishing and fuzzy reasoning. A method for fitting the results of FCM clustering is put forward by membership function and using it with attributes to achieve data discretization, in order to avoid the processing that each input data have to carry clustering operation for discretization. Using rough set theory, it can accelerate the speed of fuzzy reasoning through rules reduction with choosing closely relating attributes for commodity. Fuzzy reasoning method is also applied into commodity recommender system. Lastly, a case study conducted by the presented method is given, and the results show that this model can provide valuable advice for potential users and it is also an effective way for recommendation.