OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: A Light Rant On Ontological Commitment

Thanks for those citations, Bill.  Interesting reading. 
I recommend it not only for the insights into the 
language design but so advocates of RDF and Topic 
Maps can compare their designs to the author's advocacy 
of Prolog.  No flames, please.  It is a learning 
day every day.

However, I would place "speech acts" and other 
such classifications squarely in the domain of 
the business protocols and yes, such events 
can be used to determine the required behavior. 
It is useful to observe the behavior to determine 
if the actor is staying inside the predicted or 
contract boundaries.  Bounding named events is 
precisely why we want languages such as XLang 
and possibly the ontological services.  That 
is what a project design is:  boundaries.

That is why Tit-for-tat strategies have been 
exhaustively studied.  Simple strategies typically 
produced the highest survivor rates in simulations 
of negotiations where the Prisoner's Dilemma is 
assumed as the environmental constraint.  The author 
of your cited article alludes to this by the constraint 
"Treat this message as you do all messages of this 
type unless there is a demonstrable reason not to 
do so."   Humans do this when an RFI or RFQ comes 
in.  If we cannot understand it, we invoke a process 
to consider if it is worth considering.  Making a 
machine do this is what the machine language must 
enable and why the author of the article is investigating 
the design of such languages.

What we seem to be saying is that the first negotiation 
is to establish the negotiation pattern or strategy 
for deeper communication.  That is precisely why 
businesses issue RFIs before RFQs.   These are both 
acts of discovery to deepen the inquiry prior to the 
act of negotiation of a contract.  We negotiate the 
environment prior to instantiating controls that can 
then alter the environment itself.  We try to avoid 
the n-body problem because an environment cannot 
be managed or costed.  Consider the discussion of the 
problem of using Extreme Programming methodologies 
in fixed price projects.  Method must match management 
contracts.  The ability of the programming language 
to adequately define and then interact with the environment 
is a key consideration in choosing it as a tool.

The cultural problem, as my GE manager 
expressed it in 1989 is, "what you are talking about, Len, 
f**ks up the game".  Where the game is prized more 
highly than the potential to improve the results 
for all parties, he is right.  But times change 
and so should cultures.  Choose wisely.


Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

-----Original Message-----
From: Bill dehOra [mailto:BdehOra@interx.com]
Sent: Wednesday, January 03, 2001 6:31 AM
To: xml-dev@lists.xml.org
Cc: Bullard, Claude L (Len)
Subject: RE: A Light Rant On Ontological Commitment

> One problem is that the *intent* of language is 
> determined in the context of the culture from 
> which it emerges and within which semantics 
> evolve.  A relationship of language to culture 
> (domain to environment) is a reciprocal 
> control over the evolution of the thing(s) 
> described. We must know both what is *meant* 
> (the semantic measure within the system) 
> and the *intent* (the semantic measure of 
> the sender to receiver).  This becomes very expensive.  

Not neccessarily. Speech acts can determine intent. 

> Multi-lingual and muli-cultural are reciprocal  
> issues.  We are typically better served as 
> you point out by dealing with the transaction/contract 
> level where we can make constraints testable 
> and predictable based on observable behaviors.  

Michael Covington: <http://www.ai.uga.edu/~mc/>: "On Designing a Language
for Electronic Commerce" and "Speech Acts in Electronic Communication, KQML,
and X12", both available at the url. And Scott Moore's FLBC:

> An ontology is just a document.

An ontology can be put into document form, yes.