[Top] [All Lists]

Re: [ontolog-forum] Need advice - Request a quick opinion on ontology la

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Alex Shkotin <alex.shkotin@xxxxxxxxx>
Date: Wed, 25 May 2011 19:41:38 +0400
Message-id: <BANLkTimEPOuyC-a1XqV5Bd=o2DgnxfeLqw@xxxxxxxxxxxxxx>

when we are talking about languages and tools we should point out to real things available (ontologically speaking).
But if we talk about mathematical background beyond code and data I'd like to ask community to look at our presentation under the modest title
"Formal Representation of Knowledge in Petrology"

Semanticaly yours,


2011/5/25 John F. Sowa <sowa@xxxxxxxxxxx>
Dear Matthew, Alex, Brand, and Doug,

Before commenting on your notes, I'd like to cite the following
slides about the current state of the art and the need for a better
integration of commercial IT with semantic systems:

   Integrating Semantic Systems

The last section of that presentation makes some comments about
future directions, which I elaborated in the following article:

   Future Directions for Semantic Systems

> [Duane] also left off your list entity relationship modelling tools,
> like UML (which does much more than just entity-relationship modelling).

I agree.  UML is a good example of the way graphics can be used to
display a wide range of design patterns for specifying both declarative
and procedural information.  Furthermore the "Three Amigos" who merged
their methodologies into the "Unified" ML system did a good job of
integrating their notations with mainstream commercial IT.

My main reservation about UML is that they did not start with
a unified logical foundation from the beginning.  Since then,
Common Logic has been used to formalize the diagrams.  That's good,
but it would have been better to *start* with logic than to paste
it on as an afterthought.

> OWL 2 is the best language (Manchester OWL Syntax at first) with
> reasoners like Pellet and Protege 4(3) for ontology development.
> And if you need linear inequities (as we do) then Racer comes.

That last sentence is one more example of how and why the Semantic
Web has failed.  The UML developers set out to integrate *all* the
software development technologies in a unified methodology.  But
the SW started with nothing but syntax:  Unicode, XML, and URIs.
OWL supports a tiny part of the necessary semantics.  And as your
example shows, something else must always be added for any practical

For more detail about the limitations of the current SW tools
and some recommendations for the future, see Section 4,
pages 7 to 11, of the futures.pdf article.

I followed that link to a page with the following comment:

> Public Sector processes are very much rule-driven. Most of the software,
> developed originally to support commercial enterprise, is not rule-driven,
> or is not capable of handling 10,000s of rules in a coherent way.

I agree.  Rule-driven systems have been implemented in practical
applications since the late 1970s, and it is a shame that they have
not yet been integrated with mainstream commercial IT.

Prolog, for example, is used to support some very large, mission
critical applications, in both industry and government.  As another
example, the IBM Watson system for jeopardy uses Prolog as its
reasoning engine -- it does not use any SW tools.

> Cyc is designed to support context driven rule sets.  As for
> dynamic decision making, code would have to be written to interface
> Cyc with a dynamic I/O system.  This has been done in other cases.

I agree.  Unfortunately, Lenat has always insisted that the rules
developed in Cyc must be used with the full Cyc system.  As I
mentioned many times before, the guys who started OntologyWorks
(now High Fleet) had a better idea:  Use Cyc as a development tool
and extract the rules to run on other platforms.

Lenat did not approve of that idea.  But if he had hired the
guys who founded OntologyWorks back in 1998, they could have
made Cyc into the kind of integrated development platform that
I recommended in my iss.pdf slides.

If Lenat had done that in 1998, Duane wouldn't be asking us
for advice today.  The answer would be obvious.

Given that the integration has not yet happened, I would recommend
Prolog as the most robust rule-based language available today.
It's also an ISO standard, and it can process OWL ontologies
more efficiently than any OWL tool.


Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>