Web Services Policy Expression Alternatives

http://www.w3.org/2006/Talks/1205-ws-policy-alts/

The demand for automated or assisted Web service discovery and invocation prompted the development of the Web Services Policy framework (WS-Policy), a general purpose framework for expressing requirements, capabilities, and general characteristics for invoking a particular Web service, such as security or reliability requirements. The WS-Policy language is limited to AND and OR and functions named by QNames. Simple boolean logic allows one to discover equivalent policies and to test whether a particular connection conforms to a given policy.

While any language with AND, OR and named functions subsumes WS-Policy's expressiveness, the most interesting are those that are, intuitive, sound, and extensible. A variety of current schema (W3C XML Schema and OWL) and query languages (XQuery and SPARQL) meet these requirements to varying degrees. This paper will demonstrate and contrast expressions of Web service policies in these languages.

SPARQL and XQuery are query languages for two data different models. The current definition of WS-Policy does not presume any particular expression of service or library capabilities into any data model. All that's required is that some mechanism be able to recognize a policy and verify compliance of the available software modules. For example, this could be hard coded into an agent. Query languages like SPARQL and XQuery are designed to intuitively expression boolean logic. Using, for instance, SPARQL to express policies implies a mapping of agent capabilities into an RDF graph. Likewise, expressing agent capabilities in XML allows one to query them with XQuery.

We will describe the ways in which these languages all exceed this simple AND, OR, named function expressivity and discuss the applicability of this extra expressivity to describing Web service policies, or Web services in general. Importing this expertise from other domains informs us about potential policy expressivity. We will identify additional use cases met by adopting this additional expressivity.