Rough Sets, CCAT SPACE ontology and STEPfritz@rodin.wustl.edu (Fritz Lehmann)
Date: Fri, 16 Dec 94 16:11:21 CST
From: firstname.lastname@example.org (Fritz Lehmann)
Subject: Rough Sets, CCAT SPACE ontology and STEP
Cc: email@example.com, firstname.lastname@example.org, email@example.com, firstname.lastname@example.org
The main (if not exactly the easiest) reference
on Rough Sets is the Pawlak reference in our EGG/YOLK
paper as published in the CIKM-94 Proceedings.
My library searches and other contacts were indeed
successful. In particular I got copies of the relevant
parts of the massive STEP standard for product description.
Most of it is spatial -- extensive "data dictionaries"
with definitions of fields (data elements) in English
rather than in any formal ontology. It seems _ripe_ for
grounding in deeper ontologies of space, etc. and it
actually contains a lot of interesting ontological notions
already. A shape may be regarded as defined by a "surface
grammar", swept-form geometry, graph-theoretic boundary
"topology", finite-element, octree, etc. There is a section
on tolerances tied to other international standards.
I found someone at University of Leeds who is big in
STEP and its description language EXPRESS: Phil Spiby
at email@example.com. You should contact him
and ask him to lend you or show you the "Generic Resources"
41 (Fundamentals), 42 (Geometry and Topology), 44 (Product
Structure), 47 (Tolerances) as well as the "Application
Protocols" 203 (Configuration Controlled Design) and 224
(Process Planning Using Form Features). These are big
standards manuals -- I have them. STEP has just been
formally accepted as ISO standard 10303 which means that
commercial (non-contributing) users will have to pay a
_fortune_ for the official standard which runs to 5 or
7 thousand pages.
If you take over the SPACE subgroup of the CCAT
(Concept Catalogue/Ontologies group in the Peirce Project)
group I mentioned, your first job will be to assess
(and inspire others to assess) how the bottom-level
STEP predicates can be grounded in a lower-level general
SPACE ontology. At present the metric (versus "topological")
part of STEP is rather oriented towards Cartesian X Y
coordinates, whereas most SPACE ontology would not be.
The current blurb for the CCAT space ontology is:
3. SPACE (topological, metric, shapes, inclusions, geographic)
Walling Cyre, Serge Parkih, William Woods, Nicola Guarino
Comment: Try Randell/Cohn topology, STEP for shapes? MAPSEE
Geog. Info.System standards? Leyton? Shape grammars?
So far, nothing has been done in this subgroup.
Two research groups have already attempted to link the EXPRESS
language of STEP to Knowledge Representation languages:
Bejan & Wermeliger to Conceptual Graphs, and Van Baalen et al.
to KIF, respectively. EXPRESS is a mixture of a frame language
and a Pascal-like language, with semantics "in English".
It seems to lack nestable negations.
Note that the CCAT methodology is to exhibit all
conversation, shortcomings and ignorance on the SRKB
and Conceptual Graphs email lists (firstname.lastname@example.org
and email@example.com) until further notice. In that spirit,
I'm copying _this_ message to those lists, and you should
do the same with any CCAT message unless it's private.
Yours truly, Fritz Lehmann
GRANDAI Software, 4282 Sandburg Way, Irvine, CA 92715, U.S.A.
Tel:(714)-733-0566 Fax:(714)-733-0506 firstname.lastname@example.org