Web服务组合评分分配方法研究与实现

发布时间：2018-04-20 07:00

本文选题：Web服务组合 + 层次分析法　；参考：《昆明理工大学》2014年硕士论文

【摘要】：Web服务通过标准的Web协议提供服务给客户,将Web应用在网络间的信息交互转换为服务的交互,通过服务客户、服务提供者、服务注册机制之间一种交互式的体系结构完成服务的查找、发布和获取。随着Internet以及电子商务的快速发展,客户与企业之间以及企业内部之间的服务流程变得越来越复杂,为了完成客户的服务需求就需要将多个单一的服务组合起来,来获得一些新的功能,从而提供给客户更高的服务质量,这样在Web服务中就有了服务组合的概念,组合后的新服务称为组合服务,构成组合服务的服务称为组件服务。服务提供者为了在服务客户那获取更大的利益,往往会发布虚假服务信息,由于信息不对称,客户不能获得相关服务的真实信息,从而遭受损失。因此,在Web服务的相关研究中,Web服务信誉研究的问题受到了相关研究人员的关注,由于客户对Web服务的反馈评分是Web服务信誉度量的基础,研究客户对Web服务的反馈评分就显得非常重要。由于服务客户对于Web服务的实现细节是透明的,客户并不能区分所使用的服务是组合服务或者组件服务,在完成服务的调用后,根据服务的整体表现只给出一个满意度评分,如果客户调用的是一个包含有多个组件服务的组合服务,为了确定各组件服务的表现,就需要把服务的整体满意度评分再分配到各组件服务中,因此,关于Web服务中组合服务的评分分配方法的研究就显得非常的重要。本文首先对Web服务组合评分分配问题进行了详细的描述,然后提出了评分分配中公平的定义,并以此为研究条件进行评分分配方法的研究。为了结合组件服务的实际表现和历史表现把客户对组合服务的评分公平地分配到各组件服务,首先把Web服务流程转换为结构树流程,利用基于层次分析法的权重计算方法得到各组件服务的分配权重,再通过考虑组件服务的实际服务质量和服务提供者公告的有效区间的关系,计算各组件服务的偏差函数,同时结合组件服务历史表现计算对评分分配的贡献值,综合得到各组件服务分配到的评分值。通过实验结果验证,本文提出的关于Web服务中组合服务的评分分配方法具有良好的合理性和公平性,并且很好的满足文中所提的公平分配的条件。最后,本文完成了一个支持Web服务组合评分分配方法的原型系统,该系统采用C/S结构,使用Java语言进行开发。该系统首先需要客户从服务流程库中加载工作流,然后在加载进入系统的服务流程转换为结构树流程,根据分析出的结构树信息,得到组件服务数目,然后加载各组件服务的QoS值进行计算,得到各组件服务分配到的评分值。
[Abstract]:Web services provide services to customers through the standard Web protocol, and convert the information interaction of Web applications between networks into service interactions, through service customers, service providers, An interactive architecture between service registration mechanisms accomplishes the search, publication, and acquisition of services. With the rapid development of Internet and e-commerce, the service processes between customers and enterprises and between enterprises become more and more complex. In order to fulfill the customer's service requirements, it is necessary to combine a number of single services. In order to obtain some new functions and provide customers with higher quality of service, there is a concept of service composition in Web services. The new services after composition are called composite services, and the services that constitute composite services are called component services. In order to obtain more benefits from the service customers, service providers often publish false service information. Because of the asymmetry of information, the customers can not get the true information of the related services, so they suffer losses. Therefore, in the related research of Web services, the research on the reputation of web services is concerned by the researchers concerned, because the feedback rating of the customer to the Web services is the basis of the reputation measurement of Web services. It is very important to study customer feedback rating on Web services. Because the details of the implementation of the Web service are transparent to the service customer, the customer cannot distinguish the service used from the composite service or the component service. After the service is called, only one satisfaction rating is given according to the overall performance of the service. If the customer invokes a composite service with multiple component services, in order to determine the performance of each component service, the overall satisfaction rating of the service needs to be reassigned to each component service, so, It is very important to study the method of grading and allocating composite services in Web services. In this paper, the problem of Web service composition scoring assignment is described in detail, and then the definition of fairness in score allocation is proposed, which is used as the research condition to study the method of score assignment. In order to equitably distribute the customer's score of composite services to each component service by combining the actual performance and historical performance of component services, the Web service process is first transformed into a structure tree process. The weight distribution of each component service is obtained by using the weight calculation method based on AHP, and then the deviation function of each component service is calculated by considering the relationship between the actual service quality of the component service and the effective interval of the service provider announcement. At the same time, according to the component service history performance, the contribution value to the score allocation is calculated, and the score value of each component service assignment is obtained synthetically. The experimental results show that the method proposed in this paper is reasonable and fair for composite services in Web services, and satisfies the conditions of fair distribution. Finally, a prototype system supporting Web service composition score allocation method is completed. The system adopts C / S structure and is developed with Java language. The system first needs the client to load the workflow from the service flow library, then the service process loaded into the system is converted into the structure tree process. According to the analysis of the structure tree information, the number of component services is obtained. Then the QoS value of each component service is loaded to calculate, and the score value assigned by each component service is obtained.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】