RE: Semantics

 
Frank,
 
If I understand you correctly, you are suggesting
- the "public behavior" of WS be public, i.e., the behavior description, and
the semantic conventions needed to interpret behavior are public
- the "profile" of any specific WS, or WS providing entity could be private
or public
- the "execution agreement" between any two entities interacting using WS
can be private or public
 
I fully agree with these. Now, when I re-read AC26.2.3, it makes sense,
though it is a bit confusing.
 
D-AC026.2.3 It must be possible to characterize a service using purely
publicly observable semantics. I.e., the semantic description of a web
service should not rely on private agreements or on unobservable
characteristics of services and agents. 
 
I may suggest a change in the wording as below. Does this address your main
point?
 
D-AC026.2.3 The semantic description of a web service should be publicly
available and interpretable. 
 
Regards,

-Suresh 
Sterling Commerce   

-----Original Message-----
From: Francis McCabe [mailto:fgm@fla.fujitsu.com]
Sent: Monday, July 15, 2002 11:21 AM
To: Damodaran, Suresh
Cc: www-ws-arch@w3.org
Subject: Re: Semantics



Suresh:  



Essentially it boils down to the interactions between entities that are
owned by different people. 


Firstly, note that there is a difference between publicly observable
semantics and publicly observed communication. I.e., we are talking about
the form of the communication/descriptions etc rather than the actual
communicated information. 


If I may draw an analogy with security here: the encryption algorithm being
public benefits all users of the encryption algorithm; and actually makes
the communicated text more secure (assuming the algorithm stands up to
scrutiny of course). 


It turns out that if the very method of describing the meaning of a service
(or anything else for that matter) relies on unobservable characteristics
then you get into a lot of trouble; in particular it becomes impossible to
test for compliance. 


The wording could be clearer; but the intention is to make sure that the
principles for any semantic descriptions are set out properly. It isn't
necessary to preclude a private agreement; what is necessary is to preclude
a private agreement about the forms of messages and descriptions. For
example, we might have an out of band agreement that the words yes and no
should be swapped in their meaning. If that is not documented then everyone
is in trouble, especially you when I renage on our agreement! With a public
semantics it makes possible services such as escrow, non-repudiation etc
etc. As well as generally oiling the wheels of interactions. 


BTW, the requirements that come out of this goal should, IMO, primarily
result in requirements to the semantic web folks. From our POV we merely
need mechanisms to permit the description of services in a clear way, and to
provide architectural elements and mechanisms for managing descriptions. 


Semantics is a critical piece of the overall web services field, simply
because of the public nature of the Internet. 


Frank McCabe 



On Friday, July 12, 2002, at 04:15 PM, Damodaran, Suresh wrote: 


-----Original Message----- 

From: Francis McCabe [mailto:fgm@fla.fujitsu.com] 

Sent: Friday, July 12, 2002 5:23 PM 

To: www-ws-arch@w3.org 

Subject: Semantics 


D-AC026.2.3 It must be possible to characterize a service using purely
publicly observable semantics. I.e., the semantic description of a web
service should not rely on private agreements or on unobservable
characteristics of services and agents. 


<sd> Why? </sd> 

Thanks, 


-Suresh 

Sterling Commerce