Service description, discovery and composition are three major research topics in the research fields of Web Services and Service Science. In the past decade, enormous research effort in Web services has been spent on service description (Paolucci et al., 2003: Du et al., 2007), discovery (Ludwig & Reyhani, 2005; Paolucci et al., 2003; Song & Li, 2005; Song et al., 2009), and composition (Agarwal et al., 2005; Du et al., 2006). In order to effectively and efficiently perform Web service discovery and composition, a comprehensive service description framework is essential. There are several semantic service description frameworks proposed to provide semantic rich service descriptions, such as OWL-S (Martin et al., 2004), WSDL-S (Akkiraju et al., 2005), and WSMF (Fensel & Bussler, 2002). The main idea of these works is to build a semantic layer either on the top of or to be integrated into the WSDL documents to semantically describe the capabilities of Web services so that a software agent or other services can reason about the service’s capabilities and knows how to interact with it.
However, the problems still exist in the current study of semantic service description and discovery:
Insufficient usage context information. Usage context information of a service represents an informative surrounding in which the service is invoked or applied. Current semantic Web service description frameworks are mostly focusing on the ontology based data and capability semantics of Web services. They do not sufficiently address the usage context information of a service. Although there are some research work (Maamar et al., 2005, 2007; Medjahed et al., 2007) addressing Web services context, they mainly study the runtime environmental context, and hence help very little in allocation of the required services at the stage of service discovery. The usage context of a service includes the information about how a service is used and its relationships with other services. This kind of context information can be helpful for service users to locate their required services.
Precise service specifications. In order to locate a required service, the current service discovery requires precisely defined technical specifications for the required service, such as service input and output data types and service capabilities in WSDL and OWL-S. This kind of information is difficult for a service user to provide at the preliminary stage of service discovery, especially when the service user is not a domain expert in the required service area.
Insufficient information about inter-relationships among services. The current work inadequately addresses the inter-service relationships. A Web service needs to interact with other Web services to achieve its capabilities. If we consider each service as an isolated individual and ignore its relationships with other Web services, the efficiency of service discovery and composition for this service will decrease.
Lack of incomplete information handling. Although some of the existing work support rules in service description and composition (Martin et al., 2004; Orriens et al., 2003; Charfi & Mezini, 2004), these rules are based on monotonic logic and reasoning which are not suitable for handling incomplete information.