Re: W3C and SWS

Merry,
 
SWS is not sufficient for its purpose and may even be a wrong approach.  However, I would focus on how to make it work or how to find a new way to achieve the goal.
 
To work on the web, a registry is necessary; but a centralized service registry may not be sufficient.  A distributed registry following the DNS architecture may work better.  The match making process must based on semantics and behavior.  That is, the message schema should be irrelevant.  It is the semantics of those messages are relevant.  The matched application must be able to automatically adapt to each other's message structure.  Without such capability, Web Services can do just fine.
 
Cheers,
William Wu


----- Original Message ----
From: merrychang79 <merrychang79@163.com>
To: xuan shi <xuan.shi@mail.wvu.edu>
Cc: public-sws-ig <public-sws-ig@w3.org>
Sent: Wednesday, October 18, 2006 7:06:48 PM
Subject: Re: W3C and SWS


 
->SWS as a whole system should have at least three
components: a "shared" ontology definition on each service domain, a
centralized service registry system that enforces such "shared" ontology
definition (anyone who wants to register the service into the system, it
has to agree with the semantic definition specified in that shared
ontology definition), and a "standardized" interface to enable the
dynamic invocation. The shared ontology definition may enable the
dynamic service discovery and matchmaking through the centrailized
registry to identify certain Web services that matches the search
criteria. When the agent can identify a list of Web services, it can
invoke anyone in that list without re-programming because all such
services have a standardized interface. <-
 
Absolutely agree what the Xuan has said. If SWS is only a car couldn't be used for any reason, we have to acknowledge that it's a faillure to some extent.
 
Regards,
Merry
 
 
 
 
 
 



-----原始邮件-----
发件人:"Xuan Shi" 
发送时间:2006-10-17 23:56:51
收件人:"dhavalkumar thakker" 
抄送:" ," ,"" ,"" 
主题:Re: W3C and SWS


Dear Dhavalkumar, Drs. Martin, Klusch, and others,

First of all, I do want to apologize to all of you if what I posted
recently in the list make you feel frustrated. Although that was not my
intent, I know some words overstep the marks beyond academic discussion
when they were misconstrued publically and disrespectfully. With my good
faith to promote the research on semantic Web services, I do hope those
whom I insulted and all others could forgive my inappropriate behavior
and look forward to a more close cooperation in this community.

Let me try to elaborate what I think about semantic Web services for
your kind attention and comments. When we target the goal  of semantic
Web services as the dynamic and automatic service discovery,
matchmaking, composition and integration, SWS is a whole system like a
car. We cannot say the brake is perfect, but the car cannot run. Or we
cannot say the car runs fast and it has the best engine but it cannot
stop. Or the car has world No. 1 battery but that car is a junk. And so
on. It's the same view and logic to SWS. 

For this reason, we cannot just work on certain components of SWS, but
have to consider how to make the whole systems functions correctly and
appropriately. No matter how perfect each component is, if the car (SWS)
does not function well, it may just look like a negative junk. However,
when we see that the car functions well, this car may not have the best
components.

In my opinion, SWS as a whole system should have at least three
components: a "shared" ontology definition on each service domain, a
centralized service registry system that enforces such "shared" ontology
definition (anyone who wants to register the service into the system, it
has to agree with the semantic definition specified in that shared
ontology definition), and a "standardized" interface to enable the
dynamic invocation. The shared ontology definition may enable the
dynamic service discovery and matchmaking through the centrailized
registry to identify certain Web services that matches the search
criteria. When the agent can identify a list of Web services, it can
invoke anyone in that list without re-programming because all such
services have a standardized interface. 

Again, I apologize to all of you in this community, especially to team
members of OWL-S, WSMO, and SAWSDL, for those recent unhappy events I
generated. I welcome any kind of criticism and advice from you
publically or by private email contact. I hope my suggestion would show
some positive points for attention and discussion.

Best wishes,

Xuan




>>> "dhavalkumar thakker" <dhavalkumar@xsmail.com> 10/17/2006 5:12 AM
>>>

I think Tim Berners Lee answered the question, didnt he? 

Dear shi,

with all respect, if you find something is wrong, please suggest
something which you 
think is right, inplace of just pointing out to stuff which you think
is
wrong...
Because all we are getting from you is negative, negative and more
negative...



best regards,

Dhavalkumar










美 女 恐 怖 败 家 秀 ( 组 图 ) 
独 家 披 露 ! 小 资 女 人 8 个 绝 顶 隐 秘 的 趣 事 ( 组 图 )