Re: SUO: Universal Time, other universals, and cultural contextof SUO.

[Adam Pease <apease@ks.teknowledge.com>]

John,
   This sounds like very valuable work.  Did Iris publish the set of 
relations she used?  Is there a good summary reference on Talmy's work?

Adam

At 01:41 AM 5/18/2001 -0400, John F. Sowa wrote:

>Pat,
>
>I just remembered another important example of the relevance
>of natural language semantics to knowledge representation.
>
> > But if it really is a semantics of English, I doubt if it will be
> > of great KR interest: there are so many things in the world that
> > English has no semantics for.
>
>In 1987, I was spending several months at the IBM Palo Alto
>Scientific Center, and Gio Wiederhold suggested that I teach
>a course at Stanford, which has a slot for visiting professors
>from industry.  The course I taught was entitied "Conceptual
>Structures", and of course it used my CS book as the text.
>But I did cover many related topics, including the relationships
>between NL semantics and KR.
>
>One of the students in the class, Iris Tomlein, was working on
>her PhD in Civil Engineering, using KR techniques for representing
>spatial relationships in various kinds of structures.  She asked
>me for references on various kinds of spatial relations that might
>be suitable for representing her problems.
>
>I suggested that she look at some of the papers by the linguist
>Len Talmy, who had done a lot of work in collecting and analyzing
>spatial relations as represented in many different languages.
>Iris T. found that Talmy's primitives, which were common across
>many languages, were very useful as a basic set for kn. rep. of
>spatial relationships, and she used them as a basis for her
>dissertation.
>
>A few years later, I happened to run into a professor from the
>Stanford civil engineering dept., who said that they done a lot
>more work extending and developing that original set of relations
>that Iris T. had started with.
>
>John Sowa

Adam Pease
Teknowledge
(650) 424-0500 x571