MCC's functional interface language
Tom Gruber <Gruber@sumex-aim.stanford.edu>
Full-Name: Tom Gruber
Message-id: <2871771766-14522710@KSL-Mac-69>
Date: Tue, 1 Jan 91 18:22:46 PST
From: Tom Gruber <Gruber@sumex-aim.stanford.edu>
To: ontolingua@sumex-aim.stanford.edu
Subject: MCC's functional interface language
The MCC natural language group has specified a functional interface to
a KB that is similar to the cyc-access-functions.lisp interface for
the frame editor and simulator (funtions in the "UNITS:" package).
They have implemented connections to LOOM and Algernon.
It assumes a standard frame system plus a little bit of ontology
about events, substances, animates, etc that is needed for natural
language (like the Penman ontology). It is built on CLOS generic
functions.
They have written a draft paper that I will duplicate and put on the
shelf in 1122.
I would like to move the ontolingua in the direction of a functional
interface, making the assumptions about frameness and "content
ontology" explicit as higher order relations and model operators.
Many of the functional interface functions, such as instance-p,
get-slot-inverse, and event-p then can be viewed as efficient "HL"
shortcuts for queries and assertions that are well-defined at the
"EL" in terms of the basic ontology. The utility of specifying the
ontological assumptions in the ontoligua level is that they can be
given with a KB to describe the assumptions of an application program
that uses the functional interface. Alon, I think this might address
your need to know "what is supported efficiently" in the language as
distinct from "what is expressable".
Please take a look at what the MCC group has done and let us know
whether and how we can build on their design decisions for the
functional interface.
tom