Design Of Distributed Workflow System And Research Of Webservice Composition

Service-Oriented Architecture is getting more and more attention of the military stakeholders. But how to develop a workflow system based on the Web service is becoming one of the research focuses in Command Information System.Firstly, some research works on SOA in Command Information and Workflow system are analyzed. And the frequently used service selection or composition methods and systems are presented. Then, the generation of the Workflow is analyzed and the relevant specification and the concepts and characteristics of Web Service system are also analysed. Serveral key technologies of Web Service are also elaborated.Secondly, according to the criterions and features of the Command Information System, a framework of distributed Workflow system based on Web service is designed, inclueding process modeling tools and workflow engine. And this thesis illustrates the design method of process modeling tool detail. Namely, on the basis of the MVC framework by adding to the orientation of model, the orientation of view, agent and so on to improve the original MVC and realize remote call in the distributed environment. And local model and command continue running, but will not wait for the remote service return to block.In order to improve the efficiency of distributed workflow process modeling and automatic and intelligent level, this thesis puts forward some new service composition and algorithms. First, a service composition algorithm based on rules is proposed, which is improved based on the history case and this algorithm gets obvious improvement on tiem and space. When the historical cases of service are few, the algorithm will fail. So this thesis proposes a semantic similarity mining algorithm based on unsupervised learning which can be applied to service interface matching.Finally, according to the methods proposed above, a prototype system has been developed, of which the architecture and technologies are explained. And the basic functions of this system are demonstrated by an unmanned platforms assignment case. The rationality and feasibility of the methods and technologies in this thesis have been proved.