Re: ONT RE: Ontology case study
On Thu, May 30, 2002 at 08:33:38AM +0200, Chris Partridge wrote:
> John wishes to slake his thirst.
> John believes that by taking a drink he'll slake his thirst.
> Therefore, John takes (or decides to take, or ought to take) a drink.
>
> CP>This seems to me not good enough. One needs to distinguish between
> what I (i.e. the system) must do and what other things must do ...
Well, why not just program the system so that it carries out the
conclusion of any practical syllogism it deduces?
> - and
> also that I actually do it. Though I suppose a feature of computer
> systems (ignoring bugs etc.) is that they do what they ought to do.
Right -- taking the conclusion of a PS to be an ought statement, program
the system so that it does what it deduces that it ought to do.
> I am not arguing that this cannot be done somehow in a logical system,
> but that it is a kind of workaround. And that in a typical commercial
> system this is not - action is natural.
This point is lost on me. Why is this a "workaround"? What is being
worked around?
> > This leads onto another problem with what this discussion has
> > labelled "ontologies." The strong roots in predicate logic
> > particularly FOL. As is well known (see e.g. p. 48 of Lowe?s latest
> > book) predicate logic was developed for mathematic applications and
> > so is not well crafted for more mundane uses.
>
> That really doesn't seem to follow on the face of it. Surely it
> depends upon the "mundane" uses you have in mind; what do you have in
> mind? And I'm sure Lowe provides some arguments; can you summarize
> any of them?
>
> CP> Lowe's arguments involve distinguishing between the possession of
> properties in time in the those where time is not a concern.
OK, yes, that is a genuine concern. Why does some form of temporal
logic not answer Lowe's arguments?
> Eternal (mathematical) truths such as 2 + 3 = 5 do not seem to involve
> time. We do not say two plus three is five now - or two plus three
> was five last week.
Actually, these constructions sound just fine to me. Granted, the use
of a temporal qualifier often connotes that the statement so qualified
was NOT true at other times, but that is just a conversational
implicature.
> This is not true for the state of ones overdraft
> - which we can at least hope will at some time be zero. Anyway to
> quote "But modern logic was designed with the language of mathematics
> largely in mind, so it would not be surprising if the conception of
> predication, or property ascription, which it encourages is a
> fundamentally tenseless one.
I don't see why it follows that because modern logic happened to be
arise with respect to a domain of eternal objects that there is anything
"fundamenatally tenseless" about predication. Add some representation
of tense and predication is no longer tenseless. Granted, I'm going on
one little passage here.
> Philosophers trained in modern logic may
> feel there is something either obscure or else superficial in the
> notion of irreducibly tensed predication. They may tend, even if not
> deliberately, to model the possession of properties by time-bound
> objects on the possession of properties by abstract, timeless
> objects."
>
>CP> Note that databases were designed specifically for
> time-bound objects and actions - however ill conceived the design was.
Sure. And we can represent what goes on over time inside a DB with
standard temporal logic just fine.
>CP> For example, as far as I can see (probably not very far) SUMO does
> not have a straightforward distinction between properties that are
> time-bound and properties that are not.
That may be, though I suspect Ian would argue otherwise.
> "temporality of predication"? Definition please.
>
> CP>The attribute of time-bound properties to objects.
Again, I don't see what standard temporal logics don't give us exactly
that.
Best,
-chris