Research suggestions

Dear All,

I reviewed some samples for SWS again and have one question or suggestion
for your kind attention and advice. In the example of BravoAir
(http://www.daml.org/services/owl-s/1.1/examples.html), I think the
combination of service profile and process description is enough for users
to understand and invoke any kind of services in BravoAir either by
WSDL/SOAP or directly by HTTP/POST without the need of grounding OWL file.
This will be the same issue to WSMO. That's to say, the service requester
just needs to send a request to the provider who will process the request
and send back the response for each single request (search ticket
information, book ticket, confirm reservation, etc.) if you agree that
requester cannot send three requests at the same time.

By exchanging the service request and response in XML document, WSDL/SOAP
Web services can be easily implemented via HTTP/POST. Then it is the same
logic as the so-called REST Web services. So why do we still need such a
complex framework of WSDL/SOAP rather than the simple HTTP/POST? I know
grounding has been a big problem in SWS either for OWL-S or WSMO. However,
since we can ignore such topic from the framework of OWL-S or WSMO, I am
wondering if such research groups would like to try this method for
implementation.

In my opinion, semantic Web services can be invoked in any way without the
need of grounding elements or description files once we define the protocol
specificiations (either a WSDL interface or an HTTP server URL) inside the
service description files, such as the process.owl file. Then service
requesters will understand how to send the request to the service providers
to get the reponse. In conclusion, service description can be separated from
service implementation. Once the service can be defined and described
explicitly, it can be invoked either by WSDL/SOAP or by HTTP/SOAP by
exchanging such an XML document. In this case, why do we need WSDL 2.0? We
need more knowledge engineers and standards to share and exchange the
domain-specific, standardized service semantics.

Received on Tuesday, 15 November 2005 05:50:52 UTC