面向用户需求的Web服务发现与选择

发布时间：2018-05-03 12:32

本文选题：Web服务组合 + 服务发现　；参考：《济南大学》2014年硕士论文

【摘要】：Web服务作为面向服务体系架构(SOA)和面向服务计算(SOC)的实现，以其自描述、跨平台、松耦合等特性，已成为企业业务流程集成的一个重要机制。随着Web服务技术的发展，越来越多的企业将自己的业务流程打包成Web服务，这些Web服务在功能、质量等方面不尽相同。Web服务组合技术的出现解决了细粒度的单一Web服务难以满足复杂业务逻辑的矛盾，最大限度地提高了现有Web服务的可重用性和使用灵活性。 Web服务组合技术的关键环节是如何根据用户需求，找到能够实现其业务逻辑的候选服务集，并从中选择用以组合的最优Web服务。现有的关于服务发现的主要工作，大多都假定已给定完成Web服务组合所需的各类功能相似、服务质量不同Web服务集，基于全局或局部QoS最优的方法进行服务发现与服务选择，但对候选服务集的构造以及用户QoS偏好方面研究有所欠缺。此外，用户有时候真正关心的是Web服务的一些具体内容细节信息，服务发现的过程中应该体现这些细节信息；语义Web技术，，比如语义Web服务描述语言OWL-S，可以描述这类细节信息，但语义Web服务本身也有很大局限。针对现有工作的局限性，本文从三个方面考虑基于用户需求的Web服务发现与选择问题： (1)基于功能层面的服务发现。解决用户的业务逻辑需求可以由哪些抽象服务类实现的问题，最终目的是得到针对用户需求的候选服务集；候选服务集的建立为后续服务选择的进行提供基础。 (2)基于服务质量QoS的服务选择。建立组合QoS权重表达模型，根据用户对QoS属性的偏好和专家打分得到不同属性的权重；通过构造遗传算法，从全局的角度考虑Web服务组合的用户QoS满意度，并以此对候选服务集进行选择。 (3)基于服务内容的服务选择。提出并建立Web服务内容表达模型，通过计算服务赋值向量和用户的需求向量之间相似度的方法，确定某一Web服务是否能够满足用户对服务内容方面的要求，并以此为依据对候选服务集进行过滤。基于功能层面的服务发现形成候选服务集。在候选服务集的基础上，综合应用基于QoS和基于服务内容的服务选择，最终可得到功能层面、服务质量层面和服务内容层面最大程度符合用户需求的Web服务组合。这在总体上实现了一个Web服务发现与选择的层次模型，本文在最后讨论了该模型的利弊，并提出了另外一种可能的模型——解析模型的思路和需要解决的问题。
[Abstract]:As the realization of SOA (Service-Oriented Architecture) and SOC (Service-Oriented Computing), Web Services has become an important mechanism of business process integration for its self-description, cross-platform, loose-coupling and so on. With the development of Web services technology, more and more enterprises are packaging their business processes into Web services, which are functional, The emergence of Web service composition technology solves the contradiction that the fine-grained single Web service is difficult to satisfy the complex business logic, and improves the reusability and flexibility of the existing Web service to the maximum extent. The key link of Web service composition technology is how to find the candidate service set which can realize its business logic according to the user's demand, and select the optimal Web service to compose from it. Most of the existing work on service discovery assumes that all kinds of functions that have been given to complete Web service composition are similar, that the quality of service is different in Web service set, and that service discovery and service selection are based on global or local QoS optimal methods. However, there is a lack of research on candidate service set construction and user QoS preference. In addition, sometimes what users really care about is some specific details of Web services, which should be reflected in the process of service discovery; semantic Web technologies, such as the semantic Web Services description language OWL-S, can describe such details. But semantic Web services themselves are also very limited. In view of the limitations of the existing work, this paper considers the problem of Web service discovery and selection based on user requirements from three aspects: Service discovery based on function level. To solve the problem of which abstract service classes the business logic requirements of users can be implemented, the ultimate goal is to obtain candidate service sets for users' needs, and the establishment of candidate service sets provides the basis for the subsequent service selection. Service selection based on quality of Service (QoS). The weight representation model of composite QoS is established, and the weight of different attributes is obtained according to the user's preference for QoS attribute and the expert score. By constructing genetic algorithm, the user satisfaction degree of Web service composition is considered from the global point of view. Then the candidate service set is selected. Service selection based on service content. A Web service content representation model is proposed and established. By calculating the similarity between the service assignment vector and the user's requirement vector, it is determined whether a Web service can meet the requirements of the user for the content of the service. Based on this, the candidate service set is filtered. Service discovery based on function level forms candidate service set. On the basis of candidate service set, the Web service composition based on QoS and service content can be obtained by synthesizing the choice of service based on QoS and service content, and the function level, quality of service level and service content level can be obtained. In the end, this paper discusses the advantages and disadvantages of this model, and puts forward another possible model-analytic model and the problems that need to be solved.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】