Re: SUO: Vote on SUMO
Mehrnoush,
On a) since this is a project largely without precedent I don't see how
we could be expected to have a methodology. We do have a method that we've
explained but apparently some folks find that insufficient.
On b) we do have a structure and a language for building the
ontology. The language is SUO-KIF
<http://suo.ieee.org/suo-kif.html>. Could you explain what a
"meta-language" would be in this context?
On c) we do have an upper level, written mainly by Chris
Menzel. There's no need for any concept to be purely primitive (which in a
logical sense would mean not having a definition). In regards to whether a
concept deserves to be included in the upper level, I agree that the
boundaries are not terribly clear. However, at this early stage, spending
time worrying about which concepts to exclude seems premature to
me. Better to include anything which could reasonably be considered
general and cull out concepts at a later stage which are too specific and
put them in a separate domain specific extension to SUO.
Adam
At 12:45 PM 8/15/2001 +0430, M. Shamsfard wrote:
>I vote "NO" .
>
>Comments:
>I think although SUMO is the result of a great effort and should be
>appreciated, but there are some problems with it to be a base for a
>standard:
>
>I believe that such a standard should indicate
> a) Methodologies to build upper level and build ontologies upon
>this upper level
> b) The structure or meta language to build ontologies
> c) The content of upper level (which are a few primitive concepts)
>
>SUMO lacks (a), is weak in (b), and is intemperate in (c).
>
>In my opinion the structural ontology (first part) in SUMO is suitable
>to be
>a part of a base reference document for SUO. But the other parts contain
>some
>excess definitions which can be out of a standard upper level. Some of
>them are domain dependent which can be added to an ontology
>for some domain/applications and be deleted for others.
>
>On the other hand, as the usage of ontologies is growing and the
>knowledge
>they should contain is huge, manual development of ontologies become
>more
>expensive. As a solution, we should move toward automatic building of
>ontologies
>instead of manual construction. So the standard should indicate the
>essential (as minimal
>as possible) primitives,both concepts (categories or classes) and
>operators to
>build an ontology.
>
>Regards
>
>Mehrnoush Shamsfard
>
>Intelligent Systems Lab.,
>Computer Engineering Dept.,
>Amir kabir University of Technology
>Tehran, Iran
Adam Pease
Teknowledge
(650) 424-0500 x571