TSC was originally written in Forth on a Macintosh. You can read a bit of
it's history in my paper at
which describes the process a bit.
Basically, TSC was a frame-based system as you correctly assume here. QP
rules are also frames. In fact, in that system everything is represented
as a frame; the system was patterned after Douglas Lenat's program Eurisko.
It's agenda based (everything wraps up in tasks). QPT asks that you set
some initial conditions (in my parlance, you "set a stage") and then ask
rules to fire against those conditions. One or more rules will fire (or
else you don't have a match between the universe as described in your
rules). Each rule firing creates a new frame, called an Episode, with a
link to its predecessor (in time) by a link and a reference is made to the
rule that fired. When loops are detected, there will be more links to
priorEpisode and to mechanism (rule that fired). That's the really big
You can then query the Envisionment (graph made by rulefiring) to see what
happened. You can also use that envisionment to monitor dataflow in some
experiment and watch for "expectation failures" which mean the data changes
are not modeled in the envisionment, meaning, a new task is posed to study
that expectation failure.
Will TSC be available? Yup. Not yet, however. It's now jTSC and will be
a part of the Nexist project at http://nexist.sourceforge.net
That project is a testbed for an OHS-like system. Here, OHS refers to the
Open Hyperdocument System about which this list discusses. Source code for
Nexist is expected to go into the sourceforge CVS sometime this week.
At 02:27 PM 4/3/2001 +0100, you wrote:
> indeed. especially the mail dating from 1998 about QPT and TSC from one
>Jack Park that popped up. (awesome.)
>So how are the state transitions defined in TSC? Guessing: it sounds like a
>process is an operator on objects, and so can be defined as a something
>query + rules over a set of frame properties to create a new object. And
>you also use that process to determine whether an object has changed in a
>certain way? Guessing yes again.
>e.g. cow + slaughtering (process) => chunks of beef
>I.e. I know it was slaughtering because the mapping between cow and chunks
>of beef matches the slaughtering process operation.
>Is that how TSC works?
>Where can I get a copy of TSC to play with?
>From: Jack Park
>Sent: 4/3/01 1:58 PM
>Subject: Re: [unrev-II] Fwd: Fw: Re: [PORT-L] Goguen's Semiotic Morph
>Interesting thoughts, Eric.
>Which reminds me:
>It seems that an ontology might not, itself, be considered a modeling
>language for the *processes* going on 'out there'. Rather, an ontology
>sets the syntax and semantics for the vocabulary being used to model
>processes. A particularly valuable process modeling language, in my
>experience, is Qualitative Process Theory, as created by Ken Forbus for
>dissertation at MIT. Google those two entities (QPT and Forbus) and be
>prepared for some interesting reading.
>At 05:22 PM 4/2/2001 -0700, you wrote:
> >Matt Placek wrote:
> > >
> > > Many of the approaches to building ontologies seem to be fixated on
> > > 'present-tense' description of the nature of things.
> > >
> >Great question. It strikes me that you are asking about how to model
> >a state change in the context of an ontology. Lets take water and
> > water + heat => hot water
> > water + sufficient heat => steam
> > water - heat => cold water
> > water - sufficient heat => ice
> >Hmmm. That example introduces yet another issue: amountOf.
> >Simply adding heat to water does not predict which state
> >change occurs -- the result depends on the previous state
> >and the amount of heat.
> >But leaving the issues of quantity aside, for the moment,
> >your question revolved around a simpler state change:
> > cow + slaughter => food
> >Here there is time-based complexity, because once the
> >slaughter association occurs, the cow object ceases to
> >exist and the food object comes into being!
> >One way to model that in an ontology might be to come
> >up with a single thing that is BOTH a cow and food.
> >Let's call it "bag o'protein".
> >We might now be able to model "bag o'protein" has
> >having one or more states (cow or food), and model
> >the transitions from one state to the next.
> >On the other hand, using an ontology to do that modeling
> >is probably the wrong tool for the job. But it sure as
> >heck is an interesting question! (I look forward to other
This message is intended only for the use of the Addressee(s) and may
contain information that is PRIVILEGED and CONFIDENTIAL. If you are not
the intended recipient, dissemination of this communication is prohibited.
If you have received this communication in error, please erase all copies
of the message and its attachments and notify firstname.lastname@example.org
------------------------ Yahoo! Groups Sponsor ---------------------~-~>
Do you have 128-bit SSL encryption server security?
Get VeriSign's FREE Guide, "Securing Your
Web Site for Business." Get it now!
Community email addresses:
Post message: unrev-II@onelist.com
List owner: unrev-IIemail@example.com
Shortcut URL to this page:
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
This archive was generated by hypermail 2b29 : Tue Apr 03 2001 - 07:04:15 PDT