RE: FWD [Work in Progress on Semantics for Web Services (Advance Notice)]

Dear Dr. Sheth:

Sorry to make you unhappy but I just want to generate more discussion and
attention on such issues and problems in developing semantic Web services
other than keep silence to the known problems, such as how can WSDL-S add
semantics onto selected elements in the WSDL file for Address Finder Web
Services while may leave the others undefined, or how OWL-S/WSMO can solve
such problems. Also I don't think it's a problem for such typical case as
somebody said this WSDL file does not comply with WSDL 2.0. It's a design
issue and the same service can be implemented in WSDL 2.0 and then the
problem is still there.

Moreover, I remembered that Dr. Goodwin told me that document-based input
variables can be defined semantically by WSDL-S (maybe I misunderstood him)
but I just don't know how he did that. Given the following example, those 11
Web services that can be found at:
http://157.182.136.76/AItest/ws/AIDemo1/WebForm1.aspx have exactly the same
content of WSDL except the location of the services. Of course those 11 Web
services will perform different functions, but how can you describe the
semantics of such kind of Web services? Specifically how can you describe
the semanitcs of the geospatial features as input variables, as you know
even we have hundreds of different projection types and datums while the x,y
couple can be formatted in different ways with varied types of non-spatial
attribute fields and values. If there is nothing related to WSDL, then
trying to add semantics onto WSDL file whether in bottom-up approach or
top-town approach will just waste your time since you almost create a new
file to define such comprehensive features that are not included within
WSDL.

Again I think WSDL just provide an abstract interface that contains the
names of the service, functions, input/output variables as well as the
associated relationship generated in the process of OOP. As I discussed with
Dr. David Mark, in GIS domain, we also need to consider how to define the
order of input variables, how to define the purpose of the service and
functions as exactly the same functional interface can perform different
tasks and generate different results. Such issues will also be a normal
problem considering when we try to develop Web services for different kinds
of statistical modeling functions. 

Again, I suggest that semantic Web service research should have a common
ground for discussion. If GIS domain is too specific, then we may use
statistical Web services for testing and validating each approach. For
example, how can we describe the semantics of NISSWebSwap Web service
provided by the National Institute of Statistical Sciences (NISS). The
service semantics is describe at:
http://www.niss.org/WebServices/dg/WebSwap.html and WSDL URL is:
http://www.niss.org/WebServices/dg/WebSwap.wsdl

One more issue is that I am wondering whether OWL is suitable to describe
the semantics of Web services if the purpose of semantic Web services is to
help the services requesters (human beings) to understand the meaning of the
service while OWL is designed not for human beings but rather for machine to
process request and response. Any comments and suggestions will be greatly
appreciated. 

Many thanks again to all of your kind advice and support. I hope we can use
some more real world cases to discuss the problems in building semantic Web
services other than those intuitive and simple business models. Please
forgive me for those straightforward comments and I look forward to your
further suggestions.

Best wishes,

Xuan
 

-----Original Message-----
From: Amit Sheth @ LSDIS
To: public-sws-ig@w3.org
Cc: xshi@GEO.WVU.edu; Juggy.Jagannathan@mail.wvu.edu
Sent: 9/7/05 5:42 PM
Subject: RE: FWD [Work in Progress on Semantics for Web Services (Advance
Notice)]

Xuan:

I think you need to read the documents more carefully before you comment

with such
strong convictions.
We will limit the comments to the WSDL-S portion of your message.

"WSDL-S would only add semantics onto such WSDL elements ..."
Looks like you only read a part of the document that talks about
"bottom level annotation"..
Here is what I suggest: do a search for "top level"
(on http://lsdis.cs.uga.edu/library/download/WSDL-S-V1.html)
and see what is described following each of the matches.
[While we gave running examples for bottom level, I guess we would
add examples for top level/bottom up and top down annotations
if the document does not become too long; or we will create more 
appendices.]

One comment I sympathize with is that all of us often use simple, almost

trivial and
pedagogical examples.  But then we all want our documents to be 
readable, right?
That's why we need to use tools (such as Radiant for annotating
WSDL-S: 
http://lsdis.cs.uga.edu/projects/meteor-s/downloads/index.php?page=1)
and create complex examples.  In any case, more complete tools
and more complex use cases are coming up... shortly..
and I am sure we will have these before we go too far in adopting
any particular approach.


Amit Sheth with the WSDL-S team

Received on Thursday, 8 September 2005 02:50:09 UTC