A New Approach for Improving Collaborative filtering Recommender Systems

DOI

10.2991/mseee-17.2017.25How to use a DOI?

Keywords

collaborative filtering, user-based approach, time-weighted, dynamic multi-level, recommendation system.

Abstract

Nowadays, collaborative filtering (CF) is the most widely used method for proving recommendations in online environments, which is designed to filter large amounts of information in order to recommend to the user what they may be interested in. The user-based CF calculates the similarity between users by comparing the ratings of users to the common item, the rating generated at different times are weighted equally, but the effect of time on the user is not considered. In this paper, we propose a dynamic multi-level and time-weighted (DMLTW) collaborative filtering recommendation algorithm. Based on the original time-weighted, a new user-based time-weighted function is proposed, considering the influence of time factor on the recommendation result. At the same time, we present a positive and negative adjustment method to divide the user's similarity into different levels so as to achieve better recommendation quality. Experimental results show that our proposed method improves the accuracy of the user-based recommender systems and has a lower MAE compared to the reference method.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Zhipeng Tang
AU  - Zhengping Jin
PY  - 2017/08
DA  - 2017/08
TI  - A New Approach for Improving Collaborative filtering Recommender Systems
BT  - Proceedings of the 2017 International Conference on Material Science, Energy and Environmental Engineering (MSEEE 2017)
PB  - Atlantis Press
SP  - 134
EP  - 139
SN  - 2352-5401
UR  - https://doi.org/10.2991/mseee-17.2017.25
DO  - 10.2991/mseee-17.2017.25
ID  - Tang2017/08
ER  -