论文部分内容阅读
鉴于基于协同过滤的Web服务质量(QoS)预测存在无法应对User-Service评价矩阵过稀疏的问题,提出一种基于矩阵还原理论的QoS预测模型(HMR-QoS).首先,在现有矩阵还原模型的基础上增加了运用小幅噪声对用户评价的主观不确定性以及观测误差的建模;其次,提出利用用户及服务的上下文将原始评价矩阵切分为多个子评价矩阵,以在不显著影响评价矩阵秩分布的前提下实现并行计算;最后,界定了该机制的适定条件,并提出采用基于用户相似度的协同过滤方法补全过稀疏的子评价矩阵.仿真结果表明,相比协同过滤方法,HMR-QoS对评价矩阵的预测准确度提高了10%以上,且能有效识别出恶意评价.
In view of the problem that Web Services Quality (QoS) prediction based on collaborative filtering can not cope with the sparseness of User-Service evaluation matrix, a QoS prediction model (HMR-QoS) based on matrix reduction theory is proposed.Firstly, based on the existing matrix reduction model The subjective uncertainty and observational error of user evaluation using small noise are added. Secondly, the paper proposes to segment the original evaluation matrix into multiple sub-evaluation matrices according to the context of users and services, And finally, the suitable conditions of this mechanism are defined and a collaborative filtering method based on user similarity is proposed to make up for the over-sparse sub-evaluation matrix.The simulation results show that compared with the collaborative filtering method , HMR-QoS improves the prediction accuracy of the evaluation matrix by more than 10%, and can effectively identify the malicious evaluation.