RE: Draft charters for work on Semantics for WS

Please take a look at the following URL to a working Web service:

http://arcweb.esri.com/services/v2/AddressFinder.wsdl

For one single function "findAddress" in this WSDL file, most of the content
is meaningless to the service requesters to understand the functionality and
usage of this Web service function. Especially this WSDL contains an
embedded Web service, which cannot be identified by itself. How WSDL-S can
describe the meaning of such embedded Web services?

Take a look at the following URLs of developed Web services:

http://157.182.136.76/AItest/ws/WebService1/Service1.asmx
http://157.182.136.76/AItest/ws/WebService2/Service1.asmx
http://157.182.136.76/AItest/ws/WebService3/Service1.asmx
http://157.182.136.76/AItest/ws/WebService4/Service1.asmx
http://157.182.136.76/AItest/ws/WebService5/Service1.asmx
http://157.182.136.76/AItest/ws/WebService6/Service1.asmx
http://157.182.136.76/AItest/ws/WebService7/Service1.asmx
http://157.182.136.76/AItest/ws/WebService8/Service1.asmx
http://157.182.136.76/AItest/ws/WebService9/Service1.asmx

These Web services all have exactly the SAME content, except the location of
the service URLs. Of course they will perform differently and invoke
different functions on the server. However, if you add semantic annotations
on these 9 Web service, they may be the same and then such description is
meaningless.

Take a look at the following Web services:

http://157.182.136.51/agswsprojs/geoWebService/Service1.asmx 
http://157.182.136.51/agswsprojs/geoWebService/Service2.asmx 
http://157.182.136.51/agswsprojs/geoWebService/Service3.asmx 
http://157.182.136.51/agswsprojs/geoWebService/Service4.asmx 

WSDL interfaces are simple and all these services are atomic ones. How do
you know what is required and specified for the string data types point,
polyline, polygon, etc. used in these applications? If you want to see the
complexity of specification for them, please look at the Web site:

http://157.182.136.76/esriuc/mapviewer/Webform1.aspx

Click the radio buttons on the page and you will see what are required as
the input variables in details. How WSDL-S can "add" such specifications
onto one single input and output data type? As for the function
PolygonContained in Service4, how can you add semantics into such simple
WSDL that if user shifts the order of input polygons, they will get
completely different answer?

WSDL only contains the coding details of Web service which cannot reflect
the meaning, behavior, purpose, etc. of the service. From all above
examples, we can understand that it's more difficult to describe the service
semantics than to develop a service. WSDL-S may handle simple cases but may
not be capable to add such complex semantics into simple WSDL interface. 

For the long term goal, service description should be separated from service
development. That's to say, if we can describe it, then we know how to
develop it. For this reason, such service description can be deployed in any
way either by WSDL or not. In that case, I will see what's the role of
WSDL-S in such process.

-----Original Message-----
From: Jim Hendler
To: public-sws-ig
Cc: zhexuan.song@us.fujitsu.com; ryusuke.masuoka@us.fujitsu.com
Sent: 11/18/05 9:25 AM
Subject: re: Draft charters for work on Semantics for WS

I'm teaching a graduate seminar entitled "Advanced Semantic Web" here at
Maryland,  One of the people sitting in on the course is Zhexuan (Jeff)
Song, who works at Fujitsu Labs of America.  Jeff works on their Task
Computing project [1] which is focused on using semantics to compose Web
services, and he did a presentation last night comparing WSDL-S, OWL-S
grounding, data dictionary link and the WSDL to RDF mapping from the WSD
WG.    His slides [2] are the best discussion of this comparison I've
seen to date, and he has given me permission to post this note pointing
you to them --  for those trying to understand what is similar and
different about these approaches, and to understand some of the
vocabulary of this argument, these are a great starting place
 -Jim Hendler
p.s. if the WG is formed, I suggest Jeff's slides would be good to
include in the useful links section of their Web page.

[1] Task Computing: http://taskcomputing.org/
[2] Jeff's Slides: http://www.flacp.fujitsulabs.com/tce/WSDL-S.pdf
-- 
Professor James Hendler                   Director
Joint Institute for Knowledge Discovery                 301-405-2696
UMIACS, Univ of Maryland                    301-314-9734 (Fax)
College Park, MD 20742                       
http://www.cs.umd.edu/~hendler 
(New course: http://www.cs.umd.edu/~hendler/CMSC498w/)

Received on Friday, 18 November 2005 15:38:30 UTC