Web Service Discovery And Selection Based On Users’ Requirement

Web service, an implement of SOA (Service-Oriented Architecture) and SOC(Service-Oriented Computing), has become an important method for the integration ofenterprise work flow, due to its self-description, platform-independence and loosely-coupledcharacter. With the development of Web service technology, increasing number of enterpriseschoose to encapsulate their work flow as Web services, which differ in functions, quality ofservice, etc. The advent of Web service composition has addressed the problem that singlefine-grained Web service fails to deal with complicated work flows, and meanwhile bringsabout both reusability and flexibility.Given user’ requirements, discovering candidate Web service set and, among which,choosing the most suitable services for composition, is the main process in Web servicecomposition. Existing work about Web service discovery and selection mainly takes intoconsideration the global QoS (Quality of Service) optimum or local QoS optimum, under theassumption that candidate Web service set for composition is just available. But theestablishment of candidate Web service set is seldom referred, and moreover, users’ preferencefor QoS properties is not reflected. Sometimes, what users really concern is the content of Webservices, which should be considered in Web service discovery. This kind of information can bedescribed with semantic web technology, such as OWL-S (Web ontology language for services),but semantic Web service itself has got much limitation.To combat the limitation of existing work, this paper focuses on three aspects to addressWeb service discovery and selection:(1) Web service discovery based on service function. The aim is to determine exactly whatkinds of Web services are necessary for candidate Web service set, which acts as a foundationfor the subsequent course, say, Web service selection.(2) Web service selection based on QoS. A composite QoS model is established on thebasis of users’ preference and related experts’ experience. What follows is the construction ofGA (Genetic Algorithm), which takes a global QoS optimum aspect to deal with Web serviceselection.(3) Web service selection based on content of service. A model for content of services is proposed and established and, with this model, content of services and users’ requirement for itcan both be expressed in value vectors. Web service selection can be carried out based on thesimilarity of the two value vectors.Web service discovery based on service function helps to establish the candidate Webservice set, with which Web service selection based on QoS and Web service selection based oncontent of services are synthetically used. Eventually, the most suitable Web service selectioncan be obtained, with its functions, QoS and content best meeting users’ requirements. Viewedas a whole, we establish a hierarchical model to address the problem of Web service discoveryand selection. In the end, a discussion on both the advatanges and disadvantages of this modelis carried out; furthermore, a possible candidate for this model, say, the analytical model, isdiscussed, both about its solutions and problems to be solved.