RE: PROV-ISSUE-482: [external question] bundle IDs on insertion, context [prov-dm] from Miles, Simon on 2012-10-09 (public-prov-wg@w3.org from October 2012)

From: Miles, Simon <simon.miles@kcl.ac.uk>
Date: Tue, 9 Oct 2012 14:38:54 +0100
To: "public-prov-wg@w3.org" <public-prov-wg@w3.org>
Message-ID: <830EEE5C741ED54EAB28EBACFFC77984EEAAA8CA4B@KCL-MAIL04.kclad.ds.kcl.ac.uk>
Hello Luc,

I think the response is good in being precise, but perhaps we could add an intuitive statement

we have to
ensure that IDs don't clash in the database insertion code? or is the
understanding that IDs are only meant to be unique within a given bundle
context?

Dr Simon Miles
Senior Lecturer, Department of Informatics
Kings College London, WC2R 2LS, UK
+44 (0)20 7848 1166

Transparent Provenance Derivation for User Decisions:
http://eprints.dcs.kcl.ac.uk/1400/

________________________________
From: Luc Moreau [l.moreau@ecs.soton.ac.uk]
Sent: 09 October 2012 13:24
To: Graham Klyne; Paul Groth; public-prov-wg@w3.org
Subject: Re: PROV-ISSUE-482: [external question] bundle IDs on insertion, context [prov-dm]

Dear all,

I would like to bring this issue to a close.  I have drafted a response, available
from
http://www.w3.org/2011/prov/wiki/ResponsesToPublicComments#ISSUE-482_.28Bundles_and_IDs.29
and copied below.

Comments welcome.
Luc



ISSUE-482 (Bundles and IDs)

 *   Original email: http://lists.w3.org/Archives/Public/public-prov-comments/2012Aug/0004.html
 *   Tracker: http://www.w3.org/2011/prov/track/issues/482
 *   Group Response:
    *   PROV specifications define a notion of bundle, but do not define operations on bundles such as merge. The definition of such operations is left to implementations.
    *   The prov-constraints document defines a notion of validity in the presence of bundles. Validity is determined by checking validity of bundles, individually, irrespective of other existing bundles. For instance, the following document, containing two bundles is valid.

document
 prefix ex <http://example.org/>
 bundle ex:b1
    entity(ex:e1)
 endBundle

 bundle ex:b2
    activity(ex:e1)
 endBundle
endDocument


 *
    *   Other specifications may provide some guidance regarding this issue. For instance, the Architecture of the World Wide Web, Volume One, provides principles, constraints, and good practice notes about the use of IRIs.

 *   References:
    *   PROV validity and bundles: http://www.w3.org/TR/2012/WD-prov-constraints-20120911/#bundle-constraints
    *   Web Architecture and URIs: http://www.w3.org/TR/2004/REC-webarch-20041215/#id-resources
 *   Changes to the document: none
 *   Original author's acknowledgement:

[edit<http://www.w3.org/2011/prov/wiki/index.php?title=ResponsesToPublicComments&action=edit&section=40>]





On 09/06/2012 01:55 PM, Graham Klyne wrote:
On 06/09/2012 12:55, Paul Groth wrote:
Hi Graham,

I agree that the use of trig may cause confusion. I however think we
can show bundles in rdf just using the concept of a data file to
illustrate bundles so we would remove any cause for confusion.

Hi Paul,

This whole use of bundles is (potentially) pushing the boundaries of what RDF is currently defined to handle.

We can do as you suggest, but do we assume the graphs are merged or kept separately?  If separate, I don't think RDF says anything that constrains the interpretations used for un-merged RDF graphs.  If we assume they are, or can be, merged then what you said before about denoting same resource applies.

(If RDF Datasets were not on the horizon, I'd probably just be agreeing with you.)

#g
--


cheers
Paul

On Thu, Sep 6, 2012 at 1:26 PM, Graham Klyne<graham.klyne@zoo.ox.ac.uk><mailto:graham.klyne@zoo.ox.ac.uk>  wrote:
On 04/09/2012 17:37, Paul Groth wrote:
Luc,

I agree that this is inherited. Once you start using URIs you are
bound to their semantics which means they denote the same resource.
Furthermore, I think it would be weird for us to say anything about it
as it's treading on other specs turf.

With RDF as it stands, I agree.  And I agree about not treading on other
specifications' turf.

But an option that I don't think is entirely resolved is how bundles are
represented in RDF:  as I recall, some of the examples in the ontology document
use TRiG notation for bundles, which goes beyond current RDF but which *might*
correspond to datasets in the forthcoming version of RDF.

Unfortunately, I'm not aware of any consensus on semantics for datasets ... but
it is conceivable that URIs in named graphs within a Dataset won't be
constrained to use the same (model theoretic) interpretation (via which the
denotation is obtained) as the default graph or other named graphs.  In which
case the assumption of inheritance may be off.  All this is speculation, but I'm
bothered that we get to this stage without a clear steer from the RDF
Datasets/named graph situation.

Which I suppose may be why the provenance specs are currently silent on the
issue of how ids in different bundles are related.

#g
--


regards
Paul



On Tue, Sep 4, 2012 at 4:59 PM, Luc Moreau<l.moreau@ecs.soton.ac.uk><mailto:l.moreau@ecs.soton.ac.uk>   wrote:
Dear all,

I am trying to establish whether this question requires clarifications in
our documents or not.

Which IDs are we referring to here? Bundle IDs or IDs of entities/agents/etc
asserted in bundles?
I assume it's entity/agent/etc IDs.

I believe that both prov-dm and prov-constraints are silent about how to
interpret a given identifier
used in two different bundles.

For instance, we can write:

bundle b1
    entity(id)
endBundle

and

bundle b2
     activity(id)
endBundle

This is valid provenance (according to prov-constraints). It is not required
explicitly by prov-dm/prov-constraints
that the identifier/uri id denotes the same resource in both bundles.  But
isn't this a principle automatically
"inherited" from the Web architecture?

Thanks,
Luc



On 20/08/12 20:27, Timothy Lebo wrote:

Tracker, this is the original email from satra:

http://www.w3.org/mid/CA+A4wO=MwhCCMfaaRJwpBsfN6JCCOh_AhAkANxuP7wUhNNamFg@mail.gmail.com


-Tim

On Aug 20, 2012, at 3:16 PM, Provenance Working Group Issue Tracker wrote:

PROV-ISSUE-482: [external question] bundle IDs on insertion, context
[prov-dm]

http://www.w3.org/2011/prov/track/issues/482

Raised by: Timothy Lebo
On product: prov-dm

hi all,

if one were implementing a database storing prov bundles, would we have to
ensure that IDs don't clash in the database insertion code? or is the
understanding that IDs are only meant to be unique within a given bundle
context?

cheers,

satra






--
Professor Luc Moreau
Electronics and Computer Science   tel:   +44 23 8059 4487
University of Southampton          fax:   +44 23 8059 2865
Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk<mailto:l.moreau@ecs.soton.ac.uk>
United Kingdom                     http://www.ecs.soton.ac.uk/~lavm









--
Professor Luc Moreau
Electronics and Computer Science   tel:   +44 23 8059 4487
University of Southampton          fax:   +44 23 8059 2865
Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk<mailto:l.moreau@ecs.soton.ac.uk>
United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
Received on Tuesday, 9 October 2012 13:41:09 UTC