Research on Collaborative Filtering Recommendation Algorithm Based on Matrix Decomposition Method

DOI

10.2991/icmmcce-15.2015.41How to use a DOI?

Keywords

Matrix decomposition, Collaborative filtering, Data mining, Least squares, Personalization, Regularization.

Abstract

In order to realize the personalized recommendation of internet mass data, according to the characteristics of internet mining data set and combined with mathematical algorithms, this paper proposes a new forecasting and computing model of adding the regularization constraint and least square method based on the traditional matrix decomposition model (SVD), improving the speed and accuracy of the proposed algorithm. Matrix decomposition before and after improvement carries out experiments and results analysis with filtering recommendation algorithm, the experimental results show that the speed and accuracy of two prediction score calculation methods have some promotion after adding the regularization constraint and the least squares. After joining the regular constraints, the RMSE values obtained by MATLAB will monotonic decrease, avoiding the over fitting phenomenon and improving the calculation quality.

Copyright

© 2015, 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  - Juan Li
PY  - 2015/12
DA  - 2015/12
TI  - Research on Collaborative Filtering Recommendation Algorithm Based on Matrix Decomposition Method
BT  - Proceedings of the 4th International Conference on Mechatronics, Materials, Chemistry and Computer Engineering 2015
PB  - Atlantis Press
SP  - 205
EP  - 209
SN  - 2352-538X
UR  - https://doi.org/10.2991/icmmcce-15.2015.41
DO  - 10.2991/icmmcce-15.2015.41
ID  - Li2015/12
ER  -