Web Service Automation Supported by Lightweight Semantic Annotations

Service-oriented computing has brought special attention to service modeling, especially in connection with semantic technologies. The expected proliferation of publicly accessible services (both WS–∗ services based on WSDL and SOAP, and RESTful services — including Web APIs — based on native Web technologies) will necessitate tool support and automation. The research on Semantic Web Service (SWS) addresses this expected need with semantic technologies, creating SWS frameworks that especially address service discovery, composition and execution. As the first SWS standard, expressing a consensus of the SWS research community, in 2007 the World Wide Web Consortium produced a lightweight bottom-up specification called SAWSDL for adding semantic annotations to WSDL service descriptions. SAWSDL intentionally only defines an annotation mechanism, but does not specify any semantic model for those annotations. As such, it is intended as a point of convergence for SWS research and development. The development of SAWSDL created an empty space in need of filling by semantic models suitable for use as annotations in WSDL descriptions. To fill this space, this thesis defines WSMO-Lite, an ontology for service semantics that fits directly into SAWSDL annotations, covering functional, nonfunctional, behavioral and information semantics of Web services, which together form a basis for semantic automation. SAWSDL is, by design, specific to WSDL, a standardized Web service description technology. However, for a number of reasons, WSDL is not used to describe RESTful services, and the increasingly common RESTful services were overlooked by SWS research. There was a need for a description language aimed at RESTful services that would be lightweight and easily combine with existing practice. To tackle this need, this thesis proposes two microformats, hRESTS and MicroWSMO, which mirror WSDL and SAWSDL on top of human-oriented HTML documentation of RESTful services. Further, this thesis defines a minimal semantic service model that is an abstraction of WSDL and hRESTS, with which RESTful services can seamlessly be included in semantic processing with WSMO-Lite. To demonstrate the viability of WSMO-Lite, in this thesis we have adapted several SWS automation algorithms to this semantic model, namely algorithms for Web service discovery, ranking and composition. WSMO-Lite is intentionally lightweight, in order to smoothen the learning curve for adopters of SWS technologies. Our work is intended to serve as a basis for consolidation of SWS technologies and as an easy way of extending service-oriented and RESTful systems with semantic automation. WSMO-Lite was submitted to the W3C for consideration towards standardization, and acknowledged as a Member Submission [27]. The W3C Team Comment on the submission stated that it “is a useful addition to SAWSDL for annotations of existing services and the combination of both techniques can certainly be applied to a large number of semantic Web services use cases.”