Hi,
As a test for just how feasible a foundation ontology is, maybe
it’s worth looking at one particular “primitive” (I don’t
think I’m implying a logical primitive here, but an ontic primitive).
Most ontologies I’ve seen have, somewhere near the top of their type
hierarchy, some class whose members are physical things. What those members are
depends on what decisions were made when the ontologists set out to build their
ontology. Chris P calls these “metaphysical choices”, and as Pat H
rightly pointed out, these tend to be the cause of the ossification in large
ontologies.
My question is whether we can build an ontology covering all
these types of physical thing. I have heard there are ontologies that:
1)
Go fully 4D (b-series, perdurance theory). In this
case, anything that has a spatio-temporal extent is a member of the class
usually called “Individual” (IDEAS), “Possible Individual”
(ISO15926) or “Element” (BORO). There may be other fully-4D
ontologies with different names for spatio-temporal extent, but I’m not
aware of them. These spatio-temporal things can have states, and this is
usually how change over time is treated – e.g. I can talk about the state
that is me yesterday, or me today and make assertions that are only true for
those states.
2)
Go partially 4D – endurance theorists believe in
enduring processes, but not enduring objects. Processes can have states, but
objects cannot. Note that for a perdurant ontology (1), the process is just
another individual/element – it is the mereological fusion of all the
parts of the individuals that participate in the process. I think BFO is in
this category (though I haven’t looked in detail)
3)
Go 3D – I don’t know much about how these
ontologies work, but I’m told they constitute the majority of ontologies.
4)
3D+1 – I have absolutely no idea what this is,
but a UK Govt research centre recently rejected a proposal I put in because they
had concluded 4D wasn’t required and they were adopting a “3D+1”
approach. If anyone has the slightest idea what this is (I suspect they didn’t)
then I’d welcome some input
(this is almost certainly not an exhaustive list, but it’s
certainly large enough to illustrate there is a problem to solve)
Now, despite religious differences on the treatment of time,
it is tempting to think there is some common theme to these things. They are
all descriptions of physical stuff, after all. If so, how do they relate
? I’m not entirely sure the relationships can be accurately identified.
I think that unless this can be modelled, there is no hope
for a foundation ontology. What’s more, this is not the only issue that
would have to be cleared up. Some ontologies (I think DOLCE is one) are
intensional in their use of classification – i.e. two classes can exist
with the same members but different names. Others, such are extensional –
the class is uniquely identified by its members.
Cheers
--
Ian