Section 17.1. Semantics

17.1. Semantics

A lot of work (subsumed by the term semantic Web [FHL03]) is currently ongoing, with the objective of adding more meaning to Web content. Semantic Web technology supports the definition of the semantics of documents, such that intelligent search will extend or even substitute today's keyword matching and information retrieval techniques of locating relevant Web content. Machine-processable representations of the semantics of structured data will support automatic transformation and matching of different formats for the same piece of information.

Combining this technology with Web service technology promises to become useful for SOA in general, and Web services in particular. For example, it might be possible to automatically detect that two Web services provide the same functionality although their port types, operations, and messages are named differently. Semantic Web technology will determine that the messages of the two port types are in fact carrying the same information, and are aggregated the same way into operations. Additional semantic annotations that describe each of the two Web services are the basis of reasoning about the functionality of the operations. The combination of semantic Web technology and Web service technology is called Semantic Web service technology [SWSI], where brackets are implicitly set like "Semantic (Web service)" and not like "(Semantic Web) service."

One of the underpinnings of the semantic Web is the notion of ontology, which is a representation of the concepts (or entities) of an application domain and all of the relationships between the concepts. Annotating artifacts of the Web, or of Web services, with knowledge in terms of ontologies makes these artifacts processable via semantic Web technologies.

Semantic Web service technology considers other forms of knowledge for describing the meaning of Web services. For example, when the ordering of operations of a Web service is important, annotating it with a BPEL process model can make its use less error prone. In Figure 17-3, port type A is annotated by process model P. Based on the process model, operation o1 is the only operation available at the beginning of an interaction with an implementation of port type A. After o1 has been used, you can expect that operation o2 will send a message. After that, operations o3 and o4 will be available for use. Operation o3 is available only under certain conditions, such as if a variable named x has a value greater than 42.