Re: Question about OWL-S sequence

Quoting Drew McDermott <drew.mcdermott@yale.edu>:

> Actually, I am now willing to waffle.  My analogy was completely apt
> as an explanation for why it seemed at first to make no sense to put
> the same occurrence in two sequences.  We _have_ been thinking of the
> RDF encoding of (sequence A B C) as more or less an encoding of its
> syntax.  But since I did admit up front that we _could_ add arbitrary
> DAGs, it's clear that we could think of "occurrences" as nonsyntactic
> objects. 
> 
> Just to play with that case, here's a syntax for an arbitrary DAG:
> 
> (DAG (tag A (arf))
>      (tag B (barf))
>      (tag C (scarf))
>      (tag D (darth))
>      (tag E (earth))
>   (A -> B -> D -> E)
>   (B -> C)
>   (A -> C -> E))

O-Plan and other Edinburgh planners have used something like
that for many years.  For example:

  nodes
    1 action {fuel_vehicle ?v},
    2 action {check_vehicle ?v},
    3 action {brief_drivers ?v},
    4 action {move_vehicle ?v ?a};
  orderings
    1 ---> 4, 2 ---> 4, 3 ---> 4;

It's not always easy to read, but sequence notations can be
applied to the tags, e.g. (inventing a symtax that's not in
O-Plan):

  (sequence (parallel 1 2 3) 4)

> In fact, given this distinction, we can now bounce back to the
> original sequence notation (e.g., (sequence (arf) (barf) (scarf))) and
> reinterpret with this different convention.  We could write the DAG
> above thus:
> 
>    (parallel (sequence (tag A (arf)) (tag B (barf)) (darth)
>                             (tag E (earth)))
>              (sequence B C)
>              (sequence A (tag C (scarf)) E))

> However, I ran this by other OWL-S mavens in a recent telecon and they
> gagged.

Did they gag on the DAG idea or only on the parallel / sequence
example above?

I was thinking of something more like this:

  (tags 
      (tag A (arf))
      (tag B (barf))
      (tag C (scarf))
      (tag D (darth))
      (tag E (earth))
    (sequence A B D E)
    (sequence B C)
    (sequence A C E))

> Now the IDs don't refer to occurrences of terms; they refer (very
> indirectly) to events that occur whenever an occurrence of (an event
> corresponding to) the DAG occurs.

I suspect it's tricky to figure out exactly what the semantics
should be, but that sounds about right; but if I have a plan

   (sequence (go-to-station) (buy-ticket) (get-on-train))

it doesn't yet refer to any actual actions/events.  I might
use the same plan several times on different days, and so it
is mapped to several different sequences of actual actions/events.

Now, doesn't something like that happen anyway with OWL-S, regardless
of the syntax?  If I treat that

   (sequence (go-to-station) (buy-ticket) (get-on-train))

as OWL-S, the semantics seems to be much the same.

Thanks for all your help with this, by the way.

-- Jeff



But it seems to me

Received on Monday, 17 May 2004 11:10:57 UTC