¹School of Information Science and Engineering, Yanshan University, Qinhuangdao, China
 The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province,    Qinhuangdao, China

---

Abstract—The existing matrix factorization based collaborative recommendation algorithms have lower robustness against shilling attacks. With this problem in mind, in this paper we propose a robust collaborative recommendation algorithm based on least median squares estimator. We first propose a method of weight calculation to filter out the largest residuals by introducing the least median squares estimator (LMedS-estimator) of robust statistics, which can reduce the increment of target item’s feature vector caused by shilling attacks. Then we apply the method of weight calculation to RLS-estimator in order to realize the robust estimate of user feature matrix and item feature matrix. Finally, we develop a robust collaborative recommendation algorithm to make predictions. Experimental results on two different-scale MovieLens datasets show that the proposed algorithm outperforms the existing methods in terms of both the prediction accuracy and robustness.

Index Terms—shilling attacks, robust collaborative recommendation algorithm, least median squares estimator, reweighted least squares estimator, robustness

[PDF]

Cite: Fuzhi Zhang, Shuangxia Sun, "A Robust Collaborative Recommendation Algorithm Based on Least Median Squares Estimator," Journal of Computers vol. 9, no. 2, pp. 308-314, 2014.