W3C home > Mailing lists > Public > public-sws-ig@w3.org > May 2004

Re: [OWL-S] Process subClassOf IntervalEvent

From: David Martin <martin@AI.SRI.COM>
Date: Tue, 04 May 2004 16:59:48 -0700
Message-ID: <40982E74.6000705@ai.sri.com>
To: Austin Tate <a.tate@ed.ac.uk>
Cc: public-sws-ig@w3.org

Hi Austin -

Good points ... note that they are consistent with our directions ...

Austin Tate wrote:

> At 22:36 03/05/2004 -0700, David Martin wrote:
>> ... if "timeout" specifies a length of time that is allowed for the 
>> completion of (each execution of) a process (or control construct), 
>> then I think that does make sense.
>> But I don't think we can keep timeoutAbsolute.  I think that property 
>> was meant to specify an *absolute time* at which a process execution 
>> times out.
> I am not keen, as I have said before, about adding lots of these 
> properties of processes. 

Austin, in this thread, we were talking about 4 properties - 2 of which 
("ends" and "begins") are properties of IntervalEvent (which is in the 
time ontology imported by owl-s).  So I think those are already in line 
with your point of view on this.

Regarding the other 2 properties ("timeout" and "timeoutAbsolute"), they 
have been properties of processes (actually of ProcessComponent, a 
superclass of Process) for a long time, but a point of this thread is 
that we are *removing* one of them.  (So I think that is also in accord 
with your point of view.  You should be celebrating.  :-).


> There would just be so many that you might want 
> in terms of synchronicity, temporal delays and what not. A temporal 
> interval specification for an individual activity in the process is an 
> awful lot like a temporal delay BETWEEN activities for example. But we 
> would not want to add two separate properties.  Absolute or metric time 
> information is also the same... Oo course we want to specify it in some 
> cases, but by no means all.
> There are more fundamentally abstract ways to specify temporal 
> relationships and intervals as being information about the interval 
> itself (between 2 time points) and NOT associated directly as an 
> attribute of the activities in a process at all.  That means the same 
> relationships can hold between any time points including begin and end 
> time points of an activity.  This is much more uniform.  metric or 
> absolute temporal relationships can then be stated on those time points 
> or intervals as we wish, and the relationship to relevant begin/end time 
> points of activities is simpler to compute (often being able to use OR 
> algorithms).
> Austin
Received on Tuesday, 4 May 2004 19:58:46 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:32:45 UTC