Re: Revised daml-time.pddl

From: pat hayes (
Date: 10/31/02

  • Next message: pat hayes: "Re: Revised daml-time.pddl"
    >    [Me]
    >    >1) Yoav Shoham worked out an exhaustive taxonomy of eventuality types
    >    >    as part of his Ph.D. work.  Different types behave differently with
    >    >    regard to whether holding over an interval implies holding at each
    >    >    instant of the interval, and other such distinctions.  I'm not sure
    >    >    where to find the taxonomy, but we could ask Yoav.
    >    [Pat]
    >    Obviously I havn't been tracking this stuff closely enough.
    >    I would be very unhappy if daml-time requires us to swallow an
    >    ontology of 'eventualities'. Seems to me that there is no need to
    >    ever introduce eventualities in a well-designed ontology; they are
    >    just the sentences that are true at a time, and if one writes things
    >    properly then they can stay being sentences.
    >In spite of intense philosophical meditation, I have been unable to
    >disambiguate your proposal.
    Well it wasn't meant to be a proposal as such. We thrashed out the 
    details in emails with Jerry H. a couple of years ago.
    >Can you clarify?  For instance, does a
    >sentence's being true "at a time" cover the case where the time is an
    >  Do you intend to reify sentences or to make the time be an
    >extra argument to temporal predicates?
    In effect, the latter. But this would be done in a logic which allows 
    quantification over relations and functions (still first-order, 
    though) like CL. Then its easy to recursively translate 'eventuality' 
    talk back into essentially the same assertions made without the 
    eventualities. In retrospect, the eventualities turn out to be 
    entities corresponding to a pairing of an interval with a sentence 
    schema; but with the appropriate arguments in place, that is a 
    sentence schema; and in CL, you can just quantify over the relations 
    defined by those schema directly; no need to introduce a special 
    class of new entities. In effect, the eventualities can be seen as an 
    encoding scheme for writing 'temporal CL' in conventional FOL syntax, 
    which is fine as long as you don't think that there is anything 
    metaphysical about them.
    >In any case, daml-time can remain independent of any ontology of
    >eventualities, if some people decide they want one, or more than one.
    OK, then I will stop griping.
    IHMC					(850)434 8903   home
    40 South Alcaniz St.			(850)202 4416   office
    Pensacola              			(850)202 4440   fax
    FL 32501           				(850)291 0667    cell	          http:/	   for spam

    This archive was generated by hypermail 2.1.4 : 10/31/02 EST