Re: Changes to prov:Dictionary

On Wed, Jun 6, 2012 at 5:03 PM, Stephan Zednik <zednis@rpi.edu> wrote:

> Are you specifically worried about the possibility that other members may be
> asserted at a later time by someone else?  If this is an issue than perhaps
> you could use a system-specific extension of prov:Collection which utilizes
> a terminated ordered list.

I am not worried about this. I would consider a provenance account
initially in isolation, just like I would consider a textual log file
in isolation. Integrating provenance statements from various sources
might or might not give an inconsistent picture.


I'm worried that if I receive a provenance trace which describes the
members of a collection, then I don't know if these are just some of
the members or if they are all of the members, as the PROV standard
does not distinguish between the two. Perhaps that is a domain
specific problem, but I believed it would be rather general
requirement for a data structure like a Dictionary.

For my particular use case, I would be using incomplete membership for
the provenance of a workflow run that it is ongoing or cancelled, and
a complete membership once the engine can tell me a collection has
been completed (iteration finished) when we know for sure there is
nothing more.  This can obviously also be asserted by terminating the
iteration activity, so it's not a strong requirement for me - it just
seem like a source of disambiguousness.


> I must reiterate my agreement with Graham's point above that this need from
> this use case should not become a requirement for all collections defined in
> the standard.

I can agree if the group believes that dictionaries can still
generally be useful without knowing that it is complete or not (I
personally doubt this). The examples I have seen so far seem to assume
completeness, like members of the US Supreme Court (we know seat 5 and
7 are empty in :todays-us-supreme-court but can't state that without
completion).


> No attribute or class specialization will resolve the issue of trying to
> enforce CWA in RDF.

That's not strictly true. For instance if you claim that a dictionary
is a prov:EmptyDictionary, then it can't also have members without
making the provenance account inconsistent. (OWL does not enforce this
now, as it would require adding disjointness to prov:EmptyDictionary
to the domain of prov:memberOf, which is out of RL)


> I think in general the idea of 'completeness' is incompatible with OWA and
> should not be addressed in PROV-O.

Even with RDF Lists? Or would you consider that feature creep? I can
bend to that to finish the debate. :)

-- 
Stian Soiland-Reyes, myGrid team
School of Computer Science
The University of Manchester

Received on Thursday, 7 June 2012 09:41:27 UTC