具有多QoS约束的Web服务发现关键技术研究

发布时间：2018-06-13 10:20

本文选题：Web服务 + Web服务流程　；参考：《华北电力大学》2014年硕士论文

【摘要】：随着Web技术的迅猛发展,现代软件系统已经朝着分布式、跨软硬件平台、松耦合的方向发展。软件应用系统需要采取更加灵活的开发和配置方式以适应不断变化的环境。为了实现灵活的软件应用开发和配置方式,面向服务的体系结构应运而生。Web服务作为一个相对独立的软件功能实体不仅可以被灵活地开发、部署和调用,而且被作为下一代应用集成系统的范型,已经广泛地应用于电子商务、电子政府、企业应用集成系统中。近几年来,随着科学计算技术的飞速发展,基于Web服务的工作流管理系统已经从传统的基于任务控制流发展为面向数据的工作流。较传统的面向控制流的工作流程而言,面向数据的工作流中每个任务结点仍旧是Web服务,但是更侧重于服务之间的数据传递。因此,面向数据的工作流系统更适应用于大规模科学计算和数据分析任务。例如,基因组计算,地球科学、天文计算等。然而,当前面向数据工作流检索的研究没有充分挖掘工作流用户的特征和需求,整个检索过程需要熟悉相关计算机理论和技术,用户不得不学习一些流程查询语言来实现他们的目的。除此之外,当前的工作流检索也没有充分考虑流程中任务/服务结点的语义信息。考虑到面向数据工作流工作在一个开放、异构的分布式平台上,不同的服务提供商可以提供功能相似命名不同的服务,这为流程检索带来困难。因此,基于语义的面向数据工作流检索是解决开放环境下流程语义异构性的有效方法。最后,当前检索方法也没有考虑到用户检索的服务质量需求,如服务响应速度、服务成本、服务能耗,等等。本文将研究基于面向数据的Web服务流程发现技术。首先提出一种基于图结构的形式化表示方法以建模具有语义和可变时间约束的面向数据的Web服务工作流程,为后续研究提供基础理论模型。采用本体语言描述工作流中服务结点功能、输入输出数据类型等语义信息。语义、约束、输入输出信息等等构成了面向数据的Web服务的多种服务质量特性(QoS)。其次,提出一个用于具有多QoS约束的面向数据的Web服务流程语义检索的统一框架。该检索框架无缝地结合了Web服务流程中的语义信息和结构信息。再次,探索基于该形式化模型给出一种基于距离度量的相似度比较方法,并给出检索排序策略。最后,通过仿真实验验证本方法的有效性和高效性。本项目研究将为面向数据的Web服务流程检索提供新方法、新线索和新思路,将有助于提升海量科学计算的自动化水平,特别对于当前科学工作流研究、促进科学计算和数据分析加速发展有非常重要的现实意义和学术意义。
[Abstract]:With the rapid development of Web technology, modern software systems have developed towards distributed, cross-software and hardware platforms, loosely coupled. Software applications need more flexible development and configuration to adapt to the changing environment. In order to realize flexible software application development and configuration, the service-oriented architecture emerges as the times require. As a relatively independent software functional entity, Web services can not only be flexibly developed, deployed and called. As the next generation application integration system, it has been widely used in electronic commerce, electronic government and enterprise application integration system. In recent years, with the rapid development of scientific computing technology, workflow management system based on Web services has developed from traditional task-based control flow to data-oriented workflow. Compared with the control flow oriented workflow, each task node in the data-oriented workflow is still a Web service, but it focuses more on the data transfer between the services. Therefore, the data-oriented workflow system is more suitable for large-scale scientific computing and data analysis tasks. For example, genome computing, earth science, astronomical computing, and so on. However, the current research on data workflow retrieval has not fully excavated the characteristics and requirements of workflow users, and the whole retrieval process needs to be familiar with relevant computer theory and technology. Users have to learn some process query languages to achieve their goals. In addition, the current workflow retrieval does not fully consider the semantic information of the task / service node in the process. Considering that data-oriented workflow works on an open and heterogeneous distributed platform, different service providers can provide similar functions and naming different services, which brings difficulties to process retrieval. Therefore, semantic oriented data workflow retrieval is an effective method to solve the semantic heterogeneity of process in open environment. Finally, the current retrieval methods do not take into account the quality of service requirements, such as service response speed, service cost, service energy consumption, and so on. This paper will study the technology of Web service flow discovery based on data. Firstly, a formal representation method based on graph structure is proposed to model data-oriented Web services workflow with semantic and variable time constraints, which provides a basic theoretical model for further research. Semantic information such as service node function, input and output data type in workflow are described by ontology language. Semantics, constraints, input and output information and so on constitute the QoS characteristics of data-oriented Web services. Secondly, a unified framework for semantic retrieval of data-oriented Web service processes with multiple QoS constraints is proposed. The retrieval framework seamlessly combines semantic and structural information in Web service processes. Thirdly, a similarity comparison method based on distance measurement is proposed based on the formal model, and the retrieval ranking strategy is given. Finally, the effectiveness and efficiency of the method are verified by simulation experiments. The research of this project will provide new methods, new clues and new ideas for data-oriented Web service process retrieval, which will help to improve the automation level of massive scientific computing, especially for the current scientific workflow research. It is of great practical and academic significance to accelerate the development of scientific calculation and data analysis.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【共引文献】