On Wed, February 13, 2013 05:35, Matthew West wrote:
> ... (01)
> Chris Partridge made a plea for an architectural approach to developing a
> top ontology, considering the ontological commitments. There were two
> slides in particular that struck me.
> The first listed typical ontological choices
> that a top ontology needs to make, (02)
I would challenge the statement that a top ontology would have to make
these kinds of philosophical commitments. I think that John Sowa also
challenged this, but i'd like to explain in more depth. (03)
In most instances, the same (non-philosophical) conclusions can be made
no matter which metaphysical commitment is made. Sometimes things
need to be worded differently depending upon the metaphysics/philosophy
chosen, but that seems like an NL issue to me. If one is generating or
interpreting NL, one could use different modules for different philosophies. (04)
It would be useful to create sub-top level ontologies modeling the various
restrictions below, so that contexts that wish to commit to one of these
viewpoints can include them as well as a generic top-level ontology and
the appropriate domain ontologies. (05)
Domain ontologies that do not make such commitments can be used
by contexts that make different metaphysical commitments. (06)
> and be adopted by domain ontologies that
> use the top ontology, to ensure consistency. These are: (07)
Let me comment on these individually. (08)
> Perdurantism versus endurantism
> This makes the choice between whether e.g. physical objects
> wholly exist at each point in time, and pass through time,
> or are extended in time as well as space
> and have states that are the physical object for a period of time. (09)
This is a nice philosophical issue that has been debated for millennia.
The issue brings up the idea of timestamping assertions -- something
that often does not happen. If one wishes to state that at time T
object O has property P with value V, that is naturally a ternary
relation, if P is modeled as a predicate. To force this into binary
relations without reifying the situation, perdurantism would suggest
(during T (P O V)) while endurantism would suggest (P' (duringFn O T) V). (010)
Either reifying the situation or allowing for ternary predicates would not
force the use of one of these forms over the other. (011)
> Presentism versus eternalism
> Presentism takes the view that only things that exist now exist, and that
> things in the past and the future do not exist. Eternalism take the view
> that all points in time exist, and we can talk about historical objects as
> existing, and future objects. (012)
This is merely arguing over the meaning of the predicate "objectExists".
We can reify two predicates: "objectCurrentlyExists" and
"objectExistsAtSomeTime". We can take the logical "thereExists" to
not refer to the types of existence referred to by presentism or eternalism. (013)
> Absolute versus relative space, time and spacetime
> Absolute space-time essentially supports a Newtonian view of the world.
> Relative space-time would be the view of Einstein. (014)
Why should this be part of an ontology instead of a context? (015)
If this refers to a necessity for modeling time and space dilation at
relativistic speeds, that would be something that very few contexts
would need to model, since the Einsteinian equations become the
Newtonian ones (to enough significant figures) at "normal" speeds. (016)
Special contexts could define what happens at near light speeds,
but that should be independent from the definitions of almost
every type, relation, and individual in an ontology. (017)
Inter-timezone situations need to model that stated times are relative
to location. Again, i fail to see that this needs to be an issue of a top
ontology. (018)
It seems to me that any space model would always be relative to some
object . In a specific context, that object would be either understood
or stated (rigid Earth, Solar system, specific artifact, Milky Way Galaxy,
specific organism's body, ...). It might be nice to force it to be stated.
Within that context the space is "absolute". For inter-context reasoning,
one would need to compare what the absolute space models are for the
different contexts. (019)
> Modally extended versus unextended individuals
> Modality is about supporting what could be as well as what is. You can
> choose not to support modality, use a modal logic, or take a Possible
> Worlds approach. (020)
Again, the choice made here seems to be what predicates are chosen
to use. A top ontology could provide such predicates, but not force
knowledge bases to use them. Specific theories can be defined in mid-level
ontologies and have rules that use such predicates. (021)
I don't see that a top-level ontology needs to make a commitment here. (022)
> Materialism and non-materialism
> Materialism takes things that exist as matter or energy. Idealism is an
> alternative that considers that reality is essentially mentally
> constructed. (023)
A top ontology should not make commitments that prevent things from
being modeled, imho. Thus, unless this is an argument over the meaning
of the predicate "thingExists", a top ontology should not restrict modeled
entities from being composed of mass-energy (even if it extended to
accept time periods as entities). (024)
I'm not sure what domain (other than philosophy) would need to define
the term "reality". Again, that does not seem to be an issue for a top
ontology. (025)
> Extensionalism versus non-extensionalism - I - Universals
> Are sets/types defined by their membership, or to put it another way, can
> the membership of sets/types change or not? (026)
If by "set", you mean "mathematical set", then membership in a set can
not change. If types are defined by extent, then membership can not
change. If types are defined by rules, then for types that are defined
by rules rules whose results depend upon context, their membership
can change. (027)
The most general model would thus be non-extensionalism, but a
top ontology could define static types whose defining rules must be
context-independent. It could also define mathematical sets. (028)
Domain ontologies could define their types as general types or static
types as appropriate. Contexts that use domain ontologies could
also define general types as static within their own context. (029)
It seems that the selection of Extensional types should be a context
issue; types in a top-ontology should not be restricted to static
types. (030)
> Extensionalism versus non-extensionalism - II - Particulars
> Can two objects share their the same space (or space-time if you are a
> perdurantist)? (031)
This is a question on the definition of "spatial object". I'd suggest
that a top
ontology should include different types of spatial objects and clearly define
them. There is no need, imho, for a top ontology to restrict the types of
spatial object to one of these, although, of course, it can be quite useful
to do so in specific contexts. (032)
For example, i have a mass of soft clay. I mold it into a sphere. I then
roll the sphere between my hands to make a cylinder with rounded ends.
Then i form the cylinder into a snake with patterned back, eyes, mouth,
tongue, and rattles on its tail. Finally, i crush the snake form back into
an amorphous blob of clay. Was the sphere an object? ... the cylinder?
... the snake? ... the mass of clay? WHO CARES????
As long as my ontology can model the actions taken and the forms and
properties of the various shapes made, which of them should be considered
"objects" and which shouldn't is an immaterial (no pun intended) issue. (033)
> Topology of time - branching or linear.
> In branching temporal logics, each moment in time may split into various
> possible futures. Linear time does not have this possibility. (034)
Again, why must this be a decision for the top level ontology? (035)
If branching time (or branching space) is to be used by many domain
ontologies, then to prevent much re-inventing of the wheel, it would
be useful for the appropriate types and relations to be defined in a
top-level ontology. (036)
The choice to use (or ban the use of?) such predicates would be a
context issue. (037)
> Chris was arguing that you need to consider the order of the choices you
> make, there are natural dependencies between them. (038)
It would be useful to model the dependencies between such models
in the top ontology. That does not mean that the top ontology needs
to commit to any of these. (039)
> These are not the only choices that need to be made, (040)
Such decisions often are needed in specific contexts -- data contexts
and occasionally for theory contexts. (041)
> and the combinations may seem daunting, but in
> practice, the choices naturally cluster together. Certainly, your top
> ontology will be making assumptions about these issues. (042)
A top domain ontology need not make such assumptions, imho, if
definitional ontologies are separated from ontologies stating theories
about the domain. Theory ontologies may make such assumptions.
As you state below, if such assumptions are made, they should be
explicit. (043)
> It is always much better to know what assumptions you have made,
> than find yourself tripped up
> by them having made the choices implicitly, or find yourself with domain
> ontologies that are inconsistent from making different choices. (044)
Agreed. IF such assumptions have been made, the domain ontology
should (045)
> Together the choices you make can be considered an ontological paradigm
> that a top ontology embodies. (046)
"sub-top level" (047)
-- doug foxvog (048)
> ...
>
> Regards
>
> Matthew West
> Information Junction
> Tel: +44 1489 880185
> Mobile: +44 750 3385279
> Skype: dr.matthew.west
> matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
> http://www.informationjunction.co.uk/
> http://www.matthew-west.org.uk/
>
> This email originates from Information Junction Ltd. Registered in England
> and Wales No. 6632177.
> Registered office: 2 Brookside, Meadow Way, Letchworth Garden City,
> Hertfordshire, SG6 3JE. (049)
_________________________________________________________________
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/OntologySummit2013/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013
Community Portal: http://ontolog.cim3.net/wiki/ (050)
|