Mahdi Nasiri - Department of Computer, Science and Technology,Tehran, Iran
Behrouz Minaei - Department of Computer, Science and Technology,Tehran, Iran
Mansour Rezghi - Department of Computer, Science and Technology,Tehran, Iran

Model base collaborative filtering has been best method in recommender system. One of the best algorithms in it is matrix and tensor decomposition which have better result for rating prediction. In this paper we propose a new tensor decomposition method based on HoSVD algorithm that use time as independent dimension. Using time in recommender systems shows sequence of user interests better. Our method utilizes rating prediction based on previous ratings. Another innovation of it is time discretion using fuzzy method. Because idea of users have low difference in near time, we fuzzify discretion of time. Results show that fuzzy discretion in deed of crisp has better results.

dynamic recommender system,collaborative filtering, tensor decomposition, fuzzy discretion,sparse data