基于数据特征感知潜在因子的QoS预测方法  

QOS PREDICTION METHOD BASED ON DATA FEATURE PERCEPTION POTENTIAL FACTOR

作　　者：任利军[1] 范晓静[1] 肖志 Ren Lijun;Fan Xiaojing;Xiao Zhi(Hohhot Vocational College,Hohhot 010051,Inner Mongolia,China;School of Computer and Information Engineering,Tianjin Normal University,Tianjin 300308,China)

机构地区：[1]呼和浩特职业学院,内蒙古呼和浩特010051 [2]天津师范大学计算机与信息工程学院,天津300308

出　　处：《计算机应用与软件》2024年第9期304-313,369,共11页Computer Applications and Software

基　　金：国家自然科学基金项目(61070089)。

摘　　要：为了提升可扩展性与噪声鲁棒性,提出一种基于数据特征感知潜在因子的服务质量预测方法。从原始服务质量稀疏数据中提取密集潜在因子,检测用户和服务的邻域和噪声,在建模过程中引入了密度峰值聚类方法,实现了对服务质量数据邻域和噪声的同时检测,从而精确地表示给定的服务质量数据,实现对未知数据的高精度预测。在实际Web服务生成的两个QoS数据集上的实验结果表明,提出的方法能够有效提升预测精度和鲁棒性。In order to improve scalability and noise robustness,a QoS prediction method based on data feature perception potential factor is proposed.The dense potential factors were extracted from the original sparse data of QoS to detect the neighborhood and noise of users and services.The density peak clustering method was introduced in the modeling process to realize the simultaneous detection of QoS data neighborhood and noise.The given QoS data could be accurately expressed and the high-precision prediction of unknown data could be realized.Experimental results on two QoS data sets generated by real Web services show that the proposed method can effectively improve the prediction accuracy and robustness.

关 键 词：数据特征感知 潜在因子 服务质量 密度峰值聚类 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TM715[自动化与计算机技术—控制科学与工程]