A Multi-Criteria Collaborative Filtering Recommender System Using Clustering and Regression Techniques

Posted: 10 May 2017

Date Written: May 9, 2017

Abstract

Traditional Collaborative Filtering (CF) recommender systems recommend the items to users based on their single ratings which are used to match similar users. In multi-criteria CF recommender systems, however, multi-criteria ratings are used instead of single ratings which can significantly improve the accuracy of traditional CF algorithms. This research proposes a new recommendation method using Classification and Regression Tree (CART) and Expectation Maximization (EM) for accuracy improvement of multi-criteria recommender systems. We also apply Principal Component Analysis (PCA) for dimensionality reduction and to address multi-collinearity induced from the interdependencies among criteria in multi-criteria CF datasets. Experimental results on Yahoo! Movies and TripAdvisor datasets demonstrated that the proposed method significantly improve recommendation accuracy in case of precision.

Suggested Citation

Dalvi-Esfahani, Mohammad, A Multi-Criteria Collaborative Filtering Recommender System Using Clustering and Regression Techniques (May 9, 2017). Available at SSRN: https://ssrn.com/abstract=2965312

Mohammad Dalvi-Esfahani (Contact Author)

University of Isfahan ( email )

Azadi Square
Isfahan
Iran

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
627
PlumX Metrics