JohnS: (01)
> Pet, for example, is in the sublattice under Role, but
> it is also under Animal, since every pet is an animal. (02)
seems not so good as it supports the usual isa-overloading
that confuses. A 'pet' is not an 'animal' in the same
sense that a 'cat' is an 'animal', surely. Placing
one of these under Role and the other not shows
this. (03)
Why not: (04)
pet is the name of the second participant in
a having-pet-relationship. (05)
In English, we have some minimal but productive
morphology for such entities: e.g., (06)
employer, employee (07)
This provides more background to what it means
to place something under 'Role'. (08)
A cat is not a pet per se, it may only come to
participate in a having-pet-relationship. That is
why one can get degenerate usages: 'this is my
pet pebble'. This is degenerate because pebbles
don't participate in having-pet-relationships
very well, but people are weird. It doesn't
make any difference to a cat being an
animal that people are weird... (09)
An employee is not an employee per se, but
is some entity that some employing agency
employs and who identity criteria are
found in employment not in person. This
has been treated under qua-entities
and similar discussion of roles and
role fillers at considerable length
of course. (010)
Because a pet is ontologically founded on
participation, we know that being-a-pet
might not last that long. Some relations
are more stable than others: being-an-uncle
depends on little more than accidents
of birth. (But one can take on the role
of being an uncle, older sister, etc.) (011)
One advantage of such treatments is that
they retrieve many usages of phrases that are
otherwise sentenced to the 'metaphor
wastebasket'. But 'a pet is an animal'
is not retrieved: this is metaphorical!
It maps isa in the theory of natrual classes
to isa in the theory of roles. A pet
is at no time an animal; an employee
is at no time a person. An employee
may be filled by a person (in
some companies; by mules in others).
Otherwise one is just importing
the ability of language to ignore
this difference to ontology: in
one case it is advantageous (because
we 'know what we mean'), in the
the other it is a bug (because we
have to prove we know what we mean :-). (012)
Another advantage is that it is easier to
relate such treatments to a broader range of linguistic
expressions and in a broader range of languages.
For the recent discussion on CLCE in
different languages I'd just refer to
the 20+ year literature on multilingual
natural language generation. There we
find that event-based, neo-Davidsonian
semantics gets us much further much
quicker. So, if in doubt, consider
things in terms of events rather than
objects. I strongly suspect the same
will hold when considering the design
for controlled languages once the
final E is weakened. (013)
Best,
John B. (014)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (015)
|