一种基于矩阵填充的Web服务QoS预测算法

Web Service QoS Prediction Algorithm Based on Matrix Filling

在Web服务的选择中,用户需要对未使用过的服务的QoS进行预测。目前主要基于已有的服务历史使用信息对服务的QoS进行预测。但随着网络上Web服务数量的激增,用户-服务QoS数据矩阵是极度稀疏的,在该条件下,得到的预测结果并不准确。该文提出一种基于矩阵填充的混合协同过滤QoS预测算法。首先计算用户之间的相似度,得到目标用户的近邻集;然后采用奇异值分解法对近邻集中的缺失值进行填补,降低QoS矩阵的稀疏性;最后采用协同过滤法预测服务的QoS。使用真实QoS数据集进行实验验证,结果表明该算法具有较好的预测效果。

The QoS of Web services should be predicted before selecting the services.The history QoS data of services are usually utilized for the prediction.With the increase of service numbers,the QoS data is extremely sparse which may lead to inaccurate prediction results.In this paper,a hybrid collaborative filtering algorithms based on matrix filling is proposed.First,the similarity between users is calculated to obtain the target user＇s nearest neighbor set.Second,SVD method are proposed to fill the missing values in neighbor set for the purpose of improving the density of matrix.Finally,collaborative filtering method is used to predict the QoS of services.Wsdream dataset are utilized for experimental evaluation,the results demonstrate that our approach can achieve better performance than others.