采用矩阵分解模型的托攻击防御算法

提出一种基于矩阵分解模型的托攻击防御算法框架.首先,利用托攻击检测技术,度量用户是托用户的概率,并以此构造信任度权值矩阵;然后,将此权值矩阵引入到矩阵分解模型,以降低托用户攻击行为的影响;最后,通过求解新模型实现对用户评分的预测.实验结果表明:这类算法与其他协同过滤算法相比较,能够更有效地抵御托攻击.

Shilling Attack Defense Algorithm Using Matrix Factorization Model

A shilling attack defense algorithm framework based on matrix factorization model is proposed. Firstly, using the technology of shilling attack detection, measuring the probability of the shilling and constructing a trust weight matrix. Then, the weight matrix is introduced into the matrix factorization model to reduce the influence of the shilling attack. Finally, by solving the new model to realize the user values prediction. Experimental results show that this algorithms are more effective in resisting the shilling attacks compared to other collaborative filtering algorithms.