基于受限玻尔兹曼机的Web服务质量预测方法

Web service QoS prediction based on restricted Boltzmann machines

针对协同过滤算法无法有效处理数据稀疏的问题,提出1种基于受限玻尔兹曼机的Web服务质量(QoS)预测方法;第1阶段使用受限玻尔兹曼机模型对所有缺失的QoS值进行预测,并对原始的QoS矩阵进行填充;在第2阶段基于该QoS矩阵进行全局邻居筛选,同时将受限玻尔兹曼机引入到用户近邻的协同过滤模型中,以预测目标QoS值。研究结果表明:该方法能提高QoS预测精确度,在一定程度上降低数据稀疏对预测的影响。

Considering that collaborative filtering algorithm can not cope with data sparseness effectively,an approach for Web service QoS prediction based on restricted Boltzmann machine was proposed.In the first phase,restricted Boltzmann machine was used to predict all the missing QoS values and fill the original QoS matrix.In the second phase,global filtering was performed based on the filled QoS matrix to obtain the neighbors.Then restricted Boltzmann machine was integrated into user-based collaborative filtering model to predict target QoS values.The results show that the proposed method can improve the accuracy of QoS prediction and reduce the effect of data sparsity on prediction to a certain extent.