W3C home > Mailing lists > Public > public-prov-wg@w3.org > March 2012

Re: PROPOSALS TO VOTE ON (deadline: Wednesday 14th, midnight GMT)

From: Luc Moreau <l.moreau@ecs.soton.ac.uk>
Date: Tue, 13 Mar 2012 11:59:43 +0000
Message-ID: <EMEW3|fe065073cb83b9cb7143b76280f6b0b9o2CBxm08l.moreau|ecs.soton.ac.uk|4F5F36AF.6060905@ecs.soton.ac.uk>
To: public-prov-wg@w3.org
Hi Curt,

Some responses interleaved.

On 12/03/2012 13:52, Curt Tilmes wrote:
> On 03/09/2012 09:41 AM, Luc Moreau wrote:
>> Please express your vote for each proposal separately:
>> http://dvcs.w3.org/hg/prov/raw-file/default/model/working-copy/wd5-prov-dm-misc.html#proposal1 
>>
>
> I'm still not quite sure what this concept (tentatively called
> 'invalidation') really means.
>
> You're trying to define "end of lifetime" as "unavailable for usage",
> but that seems to be a tough thing to enforce.  What if someone else
> uses it anyway?  Can they no longer assert usages of that entity?
> What does that mean?  Is their provenance no longer consistent with
> the model?

This is not about enforcement, but about describing what has happened.

Of course, one can express descriptions that are not very meaningful, such
as an entity was used before it was generated.

We will have to add constraints stating that any usage has to preceded 
final usage/expiry.

>
>
> On 03/12/2012 06:09 AM, Luc Moreau wrote:
>> The first entity was the one at http://www.w3.org/TR/prov-dm/
>> entity(tr:prov-dm,[prov:type="working draft", ex:version="1"]) which
>> is no longer version 1, when v2 comes out at the same url.
>
> Regardless of the availability of the the content represented by an
> entity through a URL, we aren't allowed to use the same URI to refer
> to two different characterized entities. (Is that true?)  If something

Indeed, we are not allowed to do that in a given account.
See constraint and example that follows.
http://dvcs.w3.org/hg/prov/raw-file/default/model/prov-dm-constraints.html#unique-description-in-account

> changes to affect that characterization, we must create a new URI to
> distinguish the new thing from the old thing.
>
> The mere existence of tr:WD-prov-dm-20111215 doesn't preclude some
> activity 'using' tr:prov-dm (couldn't they still retrieve its content
> from someplace else?)
>
But the tr:prov-dm that you would retrieve after publishing 
tr:WD-prov-dm-20111215 would no longer be version 1 but version 2.

>
> I think this use case could better be accomodated by a distinct
> concept "prov:SupercededBy" (do we already have something like that --
> I seem to recall that coming up before?), but I still see that as an
> advisement "You *should* use the later version" -- not "You *can't*
> use the old version." that makes any assertion of a usage of the old
> version inconsistent with the model.
>

Can you expand on this suggestion of prov:SupercededBy?  How would it work?

>
>
> The second example tries even harder to make it simply impossible for
> someone to have used chicago:wkshp2002, but I still think
> disappearence of the information from the URL is a weird way to do
> that.
Again it's not about enforcement.
It is a fact that the entity at that url is no longer there today.

It would make no sense if I was stating that I used it today. I may
have used the internet archive copy of that entity, but it's a different 
entity.

>
> I guess if one of the attributes characterizing the entity with the
> URI chicago:wkshp2002 is its existence at that URL, then while the
> proceedings themselves may still exist somewhere, in some form, the
> entity represented by that particular characterization, referred to by
> that specific URI no longer does exist.
>
>
> So if I, for example, download those proceedings prior to their
> disappearance, then at some later time, perform some activity based on
> those proceedings, while my activity "used" the proceedings
> themselves, it doesn't (can't) use the entity chicago:wkshp2002, which
> no longer exists.

I would write this as follows:

entity(chicago:wkshp2002,[prov:type="workshop talks"])
activity(ex:download)
entity(ex:wkshp2002copy)
wasGeneratedBy(ex:wkshp2002copy,ex:download)
used(ex:download,chicago:wkshp2002, t1)

wasInvalidated(chicago:wkshp2002, t2)

activity(ex:otherAction)
used(ex:otherAction, ex:wkshp2002copy, t3)

where t1 < t2 < t3
>
> Too weird.
>
>
> Curt
>

Luc
Received on Tuesday, 13 March 2012 12:00:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 13:06:58 GMT