| Summary: | 博士 === 國立中山大學 === 資訊管理學系研究所 === 100 === Although the benefits of reusing services have been widely acknowledged, composing existing services to realize different user requirements is still a difficult and time consuming task. There have been research works on workflow templates and service design patterns that offer a higher level of reuse and assist in the composition process. Yet existing workflow templates are too specific and service design patterns may be too abstract and their effectiveness in assisting the composition process may be limited. In this dissertation, we propose a comprehensive service pattern model that is more flexible than existing workflow/service templates while allowing systematic instantiation into some concrete workflows. The service pattern supports flow control, abstract services, and prospect services (similar to a node in a design pattern). Also, we develop a set of reasoning rules to automatically verify and determine the assignment of instantiation parameters in service pattern. In addition, recent research into automated Web service composition focuses on control flow specification and enforcement. Traditionally, message mismatch between Web services appears as an orthogonal problem that can be solved using XML-based query languages. This study instead considers Web service composition and message manipulation holistically to support greater optimization of Web service composition. A proposed model, the WS-data model, focuses on data exchanges in composing Web services. Several operators with varying properties can be used to compose Web services and manipulate their input and output data. Experiments conducted on the Amazon Cloud platform show that these operators’ properties can be used to develop a more efficient way for realizing a complex task, expressed according to the proposed WS-data model. It is hoped that the proposed models and automated reasoning of instantiation parameters assignment will realize the benefits of SOA in the real business settings.
|
|---|
