Pat, Doug, Chris, Matthew -
FWIW, the way Roles are handled in
COSMO (an OWL ontology at present) has these characteristics:
'Role' is a top-level first order type
(class) not disjoint with anything but Attributes. This can be
a parent class of any other type.
Thus one can have, for example,
'HumanRole' which is a subtype of 'Person' and 'Role'. (for
example, 'Student' or 'President' or many others). Then when
one asserts that a 'HumanRole' participates in some action,
there is both a RoleFiller (human) and a 'Role" involved. The
logical implication is that the Person who plays that HumanRole
is the participant. Each instantiation of Role has a starting
and ending time.
There are also relations that relate the
individual that plays a Role with the Role. Most Roles will not
have specific subtypes that combine the Role and its Filler in
one type. 'HumanRole' is used because it is so important in
everyday situations.
There is also a RoleSituation, a subtype
of 'PersistentState' (same as a 'state' in many terminologies).
This has some Role filler that is the subject of that
situation. One could (COSMO does not at this point) define a
relation specifying how well that role filler fills that role in
that specific RoleSituation.
The point is that, although 'Role' is
not itself a Physical concept, it can have subtypes that are
physical objects. This may differ from some ontologies (I think
in DOLCE Role cannot have physical subtypes). For compatibility
with those ontologies, it might be necessary to include a
subtype of 'AbstractRole' that corresponds to Roles that cannot
have PhysicalObjects as instances.
The rationale for this choice was to
keep COSMO as close as possible to linguistic usage while
minimizing ontological assumptions. If one says the 'The Bishop
sent a letter' and if "Joe Smith' is the Bishop, the logical
implication is that Joe Smith sent the letter. And vice-versa,
one can say that 'Joe Smith' sent the letter, and the logical
implication is that the Bishop sent the letter. This means that
if a Role performs an action, it is not necessary that that
actions be performed **specifically** in the capacity of that
role (the Bishop can send a personal letter).
'Part' is a subtype of 'Role' and every
'Part' will be filled by an entity of some other type. Thus in
this treatment, Pump101 would be a subtype of Role and also a
subtype of 'Pump' (a physical device). The physical objects
that fill that Role at any time could also be referred to by the
role name ('Pump 101 has failed'). The individual object that
fills that Role may be identified, but it does not have to have
a separate representation other than as 'Pump101'. If it is
separately instantiated (as most manufactured components would
be), then it would be related to 'Pump101' by a role-filler
relation, delimited by the starting and ending times it plays in
that Role.
The thing that sits on the floor and plays the role of Pump101 does
have a separate identity. It probably has a make, model, serial
number, a purchase history, possibly a usage history - was Pump 209
for a while, was Pump 390 for a while, sat in the parts warehouse
for 3 months and is now Pump101. It has a BOM with lots of other
parts.
Thus far in this discussion I have not
noticed any requirements that the COSMO treatment would not
accommodate. Perhaps there are others?
Pat Cassidy
MICRA Inc.
908-561-3416
cassidy@xxxxxxxxx
---------------
[Doug Foxvog]
> Pat,
> Makes sense to me.
> May we include an attribute on role
signfying
effect-iveness? Role
> effectiveness will partially depend
on actor (if you
have ever seen
> Peter Sellers play Hamlet).
> Jack
Ontologize the role playing as an object
in its own right
-- a Situation.
This would be a Davidsonian approach. Then
you could make
whatever assertions about it are needed.
[ChrisPartridge]
Pat,
It seems to me as if you are just playing
with names
here. If you want to call it a pump *role*, that is fine. But
that what you are
describing seems not to have the qualities that many people
expect to be
essential to roles.
These (like qua entities) do not have an
individual identity
and they do not do things, they are not agents. Whereas, for
example,
spatio-temporal entities come bundled with identity. What have I
missed?
So the Hamlet example would better be Jonathon Pryce's 1992 Hamlet. Or even
better if
we use Chairman (President, Bishop or Monarch) , the difference
between
Chairman and the Chairman of Goldman Sachs.
Also, not clear to me why you cannot kick
your roles -
as, again, they are spatio-temporal entities? When Ronnie Reagan
was shot,
people said they shot the President of the US, didn't they? They
did not say
thank goodness they only shot Mr Reagan - they could not shoot
the President as
he is a role.
Regards,
Chris
[Pat Hayes]
Very good question, Matthew. Let me try
out an idea on
you. Your P101 is actually a role played by a pump, rather than
a pump itself.
Think of it as being like Hamlet, as played by Lawrence Olivier
(P101 as played
by S3556). You can change actors, and Hamlet is still Hamlet -
same role - and
while Olivier is playing the role, he *is* Hamlet, at least in a
sense. But
this second "is" cannot be identity, since you can kick the
actor,
but you can't kick a role.
Both a pump and a pump-role are
spatiotemporal entities,
but they have different identity conditions. The identity of a
pump, like any
other physical object, is determined by the disposition of
pieces of material
stuff (metal, plastic, rubber), but the identity of the role is determined by its interfaces to
the rest of the system (being connected to this pipe in this
place and operated
by this controller, etc..)
You can identify a pump-phase (temporal
slice) with a
pump-role-phase, but you must not identify the actual
individuals, so its safer
to actually have a relation of 'functioning as' of the like to
attach a
role-playing thing to its role. Or, you can treat the role as a
time-dependent
property of the physical thing, but you will probably need a
CL-style ability to
have properties of properties if you go that (elegant) route.
Make sense?
Pat
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012
Community Portal: http://ontolog.cim3.net/wiki/
