Information Technology Reference
In-Depth Information
different semantics of usable languages relate to one another, it is impossible
to formally define requests, queries, or notions of a “match” between service
requests and service descriptions.
With respect to nonfunctional properties of services, WSDL-S points to
several existing proposals for standards from the Web service community. For
instance, the authors of WSDL-S remark that is is necessary to “[...] investi-
gate how to represent QoS assertions using ontologies and rules by extending
the WS-Policy framework” [1]. Likewise, as for behavioral descriptions or in-
terface descriptions, similar extensions of BPEL or WS-CDL are mentioned
as being possible. The concrete integration with such existing standards has
not yet been tackled, and remains open for future work. In terms of indus-
trial adoption of these standards, the WSDL-S team definitely has a point
here. In fact, the WSMO and WSMX working groups are investigating closer
alignment with industry standards, and the adoption of WSDL-S/METEOR-
S itself as a possible future grounding formalism for WSMO has even been
considered [131].
8.4 Summary
As we have learned in this chapter, apart from WSMO, several other ongoing
initiatives are progressing towards the development of frameworks for adding
semantic annotations to Web services.
We also learned that despite some fundamental conceptual differences,
there are several features common to these approaches. For instance, they all
seem to agree on a minimal subset of necessary aspects of semantic service
annotations:
1. General
service classifications
are common to all approaches.
2. A notion of
preconditions and postconditions
for successful service execu-
tion is common to all approaches, on the level of either atomic operations
or complex service interactions.
3. As for the
behavioral/protocol
descriptions of a service, WSMO, OWL-
S, and SWSF all allow one to encode complex behavior and interaction
patterns; in WSDL-S, this can probably be simulated to a limited extent
by adding preconditions and effects to individual WSDL operations, for
example by embedding WSDL-S into BPEL4WS [4]
4. Nonfunctional aspects (such as quality of service, cost, and availability)
can be modeled in most approaches. WSMO, OWL-S, and SWSF all pro-
vide extensible sets of nonfunctional properties in their ontologies, whereas
WSDL-S sees this as out of its scope and instead points to related WS-*
standards.
However, WSMO has some distinct features compared with the other
approaches mentioned. First, mediators are a unique top-level concept in
WSMO, whereas in OWL-S and SWSF the aspect of resolving heterogeneities
Search WWH ::
Custom Search