一种基于用户在线行为分析的Web服务动态QoS评估方法  

A Dynamic QoS Evaluation Method for Web Services based on Online Behavior Analysis of Users

作　　者：王瑞祥 魏乐 常钰 段燕飞 WANG Ruixiang;WEI Le;CHANG Yu;DUAN Yanfei(College of Software Engineering,Chengdu University of Information Technology,Chengdu 610225,China)

机构地区：[1]成都信息工程大学软件工程学院,四川成都610225

出　　处：《成都信息工程大学学报》2021年第1期51-61,共11页Journal of Chengdu University of Information Technology

基　　金：四川省应用基础研究资助项目(2018JY0506);四川省教育厅青年基金重点资助项目(16ZA0208)。

摘　　要：随着以计算机为中心的Web服务提供方式向以用户需求为中心的方式转变,针对以往动态QoS模型未能充分考虑用户在线行为等因素,导致服务质量的评估缺乏实时准确性,文中通过考虑用户在线行为所造成的Web服务流行度失衡及流行热度的区域性变化等因素,对服务流行性的评估建立算法模型,采用Logistic增长曲线和种群增长速率曲线的思想描述用户在线行为变化,并基于遗忘算法提出了服务流行热度值评估算法,能够根据单位时间内服务请求数量增长率的变化趋势,来自适应调整流行性的数值,又通过聚类算法对服务请求的热点位置密度进行划分,解决了流行热度的区域性变化问题。而后将流行性作为动态QoS模型的一项重要指标,并对模型进行重构与调整,通过建立数据误差评估、归一化和动态加权的方案实现了QoS属性的动态合成,最后通过实验验证了该模型可行有效。As the computer-centric web service provision method is changing to the user-centric web service provision method,considering that the previous dynamic QoS model fails to fully consider the online behavior of users and other factors,which leads to the lack of real-time accuracy of service quality assessment.For this reason,this article considers factors such as the imbalance in the popularity of web services and the regional changes in popularity caused by user online behaviors,establish an algorithm model for the evaluation of service popularity,the idea of Logistic growth curve and population growth rate curve was used to describe the changes of users'online behaviors.A nd based on the forgetting algorithm,the service popularity heat value evaluation algorithm is proposed,which can adaptively adjust the popularity value according to the change trend of the growth rate of the number of service requests per unit time,the clustering algorithm is used to divide the hotspot location density of service requests,which solves the problem of regional changes in popularity.Then the popularity is regarded as an important indicator of the dynamic QoS model,and the model is reconstructed and adjusted,through the establishment of data error evaluation,normalization and dynamic weighting,the dynamic synthesis of QoS attributes is realized.Finally,experiments verify that the model is feasible and effective.

关 键 词：WEB服务 服务质量 流行热度 在线行为 QoS建模 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]