基于运行日志分析的服务组合优化的Petri网方法

发布时间：2018-04-07 00:02

本文选题：Petri网　切入点：服务组合　出处：《济南大学》2016年硕士论文

【摘要】：随着互联网技术和电子商务的迅猛发展,Web服务作为信息服务资源的具体实现之一得到了更为广泛的使用。Web服务因其跨平台、松耦合的特点,支持在不同语言,不同平台上进行有效的Web服务组合,以分布式计算的模式为人们提供更为方便和丰富的服务。如何将已有的Web服务进行有效组合一直是服务计算领域的热点问题,其中包括服务选择,服务组合建模,服务组合动态执行分析等问题。Petri网因其具有严格的数学定义和适合描述并发系统的特点,近年来在Web服务组合相关的问题研究中得到广泛应用,尤其在服务组合模型的构建,服务组合中服务的行为分析等方面。本文面向Web服务组合,基于服务组合的Petri网描述,从对具体服务组合执行过程日志的分析,研究对已有服务组合模型进行优化的方法;同时基于运行日志分析,研究实际运行中服务间存在的关联关系,为服务选择提供有效的选择依据,提高服务组合的效率。本文首先从日志文件的获取和分析方法入手,设计并实现了基于Petri网的服务组合执行引擎原型系统平台,在此平台之上可以实现Web服务的注册,Web服务的组合,服务组合模型(mx Graph)与相应Petri网模型间的转换,服务组合的执行。该平台基于服务领域功能体系的OWL(Ontology Working Group)文档,解析Web服务的WSDL(Web Services Description Language)文件,将Web服务注册到本地,根据平台中展示的领域服务进行服务的组合,并将其中各类服务绑定对应的具体服务,将服务组合实例提交执行引擎执行,最后返回执行结果,同时得到相应服务组合执行的运行日志。其次,以服务组合执行日志为基础,挖掘运行日志所反映出具有强关联的抽象服务和强关联的具体服务,并且设计相应的关联服务库,把这些关联的服务存储到关联服务库中,按照关联强度大小,把抽象的关联服务推荐给用户,作为指导用户的选择,在进行抽象服务实例化时也是按照关联强度从大到小的原则进行具体服务的选择,并通过具体实验来验证由日志中获取的服务间关联的有效性及可靠性,优化了服务选择。最后,从服务组合模型的角度对服务组合进行优化。服务组合完成之后,基于运行日志,通过对Web服务组合中各服务执行时间的评价,对服务组合的Petri网模型进行优化,以实现对服务组合模型优化,提高服务组合的执行效率,有效降低服务器端系统资源消耗。
[Abstract]:With the rapid development of Internet technology and electronic commerce, Web services, as one of the concrete realization of information service resources, have been more widely used. Web services are supported in different languages because of their cross-platform and loosely coupled characteristics.Effective Web service composition based on different platforms provides people with more convenient and abundant services with distributed computing mode.How to effectively compose existing Web services has always been a hot issue in the field of service computing, including service selection, service composition modeling.Service composition dynamic execution Analysis. Because of its strict mathematical definition and suitable for describing concurrent system, Petri net has been widely used in the research of Web service composition related problems in recent years, especially in the construction of service composition model.Analysis of service behavior in service composition.This paper aims at Web service composition, based on Petri net description of service composition, from the analysis of the execution process log of specific service composition, studies the method of optimizing the existing service composition model, at the same time, based on the analysis of running log,In order to provide an effective basis for service selection and improve the efficiency of service composition, this paper studies the relationship between services in actual operation.This paper starts with the method of obtaining and analyzing log files, designs and implements the prototype system platform of service composition execution engine based on Petri net, and can realize the composition of Web service registration based on this platform.The transformation between the service composition model and the corresponding Petri net model, and the execution of the service composition.The platform is based on the OWL(Ontology Working Group document of the service domain function system, parses the WSDL(Web Services Description language file of the Web service, registers the Web service locally, and performs the composition of the service according to the domain service shown in the platform.The service composition instance is submitted to the execution engine for execution, and the execution result is returned. At the same time, the running log of the corresponding service composition execution is obtained.Secondly, based on the service composition execution log, mining the abstract service with strong association and the concrete service with strong association reflected in the running log, and designing the corresponding associated service library to store these associated services in the association service library.According to the intensity of association, the abstract association service is recommended to the user as the choice of guiding user, and the specific service is chosen according to the principle of association strength from large to small when instantiating the abstract service.The validity and reliability of the correlation between the services obtained from the log are verified by experiments, and the service selection is optimized.Finally, the service composition is optimized from the point of view of service composition model.After the service composition is completed, based on the running log, the service composition Petri net model is optimized by evaluating the execution time of each service in the Web service composition, in order to optimize the service composition model and improve the execution efficiency of the service composition.Reduce server system resource consumption effectively.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09;TP301.1

【参考文献】