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

Re: alternate example in primer

From: Luc Moreau <l.moreau@ecs.soton.ac.uk>
Date: Mon, 19 Nov 2012 10:47:39 +0000
Message-ID: <EMEW3|9c396c9548bf77106ea21e6d47f38130oAIAlf08l.moreau|ecs.soton.ac.uk|50AA0E4B.4080508@ecs.soton.ac.uk>
To: "Miles, Simon" <simon.miles@kcl.ac.uk>
CC: James Cheney <jcheney@inf.ed.ac.uk>, Ivan Herman <ivan@w3.org>, W3C provenance WG <public-prov-wg@w3.org>
Hi Simon,

I think the origin of my problem with the primer example is that it does 
not indicate
which is the common thing.  This is left to the reader to guess: I 
failed to guess
what it was.  Your suggestion of back-up file may well work. I would 
need to read it.

Thanks,

Luc

On 11/19/2012 10:35 AM, Miles, Simon wrote:
> Hi Luc,
>
> I agree that sharing types does not imply that two entities are alternates. However, I don't see the connection to the primer example.
>
> Yes, the Mercurial example is one where files are copied and the copies are normally considered to be the same file. Another, simpler version, is where I copy a file as a backup - the copy must be the same file otherwise it would not make sense to restore the original with the copy. Another example is when I copy myDocumentV1 to myDocumentV2, ahead of editing the latter as the new version of the same document.
>
> I understand that you can find examples of copying where the copies are not considered to be presenting aspects of the same thing. In all cases, it is a modelling decision, specific to the application, as to whether there is a common thing that two entities are perspectives on. There is no absolute notion of whether two entities are alternates or not, just more or less intuitive uses of the relation. That is why the primer currently says "... we may want to say that (according to our model) these are both the same file, just in a different location."
>
> To address your comments while keeping the primer as simply explained as possible, I suggest that I add to the example a statement that the copied file is to be a back-up of the original. Would that address your concerns?
>
> thanks,
> Simon
>
> Dr Simon Miles
> Senior Lecturer, Department of Informatics
> Kings College London, WC2R 2LS, UK
> +44 (0)20 7848 1166
>
> PrIMe: A Methodology for Developing Provenance-Aware Applications:
> http://eprints.dcs.kcl.ac.uk/1382/
> ________________________________________
> From: Luc Moreau [L.Moreau@ecs.soton.ac.uk]
> Sent: 15 November 2012 00:20
> To: Luc Moreau
> Cc: James Cheney; Ivan Herman; W3C provenance WG
> Subject: Re: alternate example in primer
>
> Hi james and Simon ,
>
> In a discussion today, we came up with the following example.
>
> Consider two entities e1 and e2 both instances of class Foo. Are they alternate? Could the thing they have in common be the aspect that they belong to the same class Foo?
>
> If this is an example compatible with our definition, then any pair of entities would be alternate, since they are of type owl:Thing.
>
> If this is example is not compatible with the definition, then we should avoid the example in the primer, since it seems that the common aspect of the two files is a type.
>
> James example with the hg repo is good.
>
> Another, inspired by the dm, is a resource accessed by a mobile device and a desktop cimputer. The two representations would be alternate.
>
> Professor Luc Moreau
> Electronics and Computer Science
> University of Southampton
> Southampton SO17 1BJ
> United Kingdom
>
>
> On 14 Nov 2012, at 06:22, "Luc Moreau" <l.moreau@ecs.soton.ac.uk> wrote:
>
>> Hi James,
>>
>> I am far more at ease with the example you are now giving, where there is a file in a mercurial repository,
>> and two local copies on your machine and mine, which are alternate.
>>
>> What I struggle with is a copy operation taken in isolation (outside the context of a mercurial checkout). I don't
>> see what the common thing is. It becomes something like a file with a given content ... which looks more like
>> a type description than a thing.
>>
>> If we were keeping the example of copy in the context of mercurial, with a file in the repository as the common
>> thing, than I wouldn't have any objection on this example.
>>
>> Luc
>>
>> On 14/11/2012 06:14, James Cheney wrote:
>>> Well, for example, my computer has a copy of the hg repository.  So does yours (I assume), and there is also a copy on the W3C hg repository.  All three have (copies of) prov-dm at various times.
>>>
>>> One could say that all three files (and their previous, local versions) are alternates of one common thing, "prov-dm in the main HG repository and on Luc and James' computers".  Or, one could say that only the instances of prov-dm on a single repository are alternates.
>>>
>>> I'm not saying it's particularly "clean", but there is nothing in the constraints that rules this out.  Moreover, the web architecture (AFAIK) doesn't make strong commitments to what resources really are or where they are stored.  Ultimately, this may limit how useful alternateOf is, if everyone interprets it differently.  but in that case, at least the negative consequences are limited since we don't use it for much - it is essentially documentation of what users/systems believe about how different entities are related, that is, what the underlying "things" are.
>>>
>>> --James
>>>
>>>
>>> On Nov 14, 2012, at 11:08 AM, Luc Moreau wrote:
>>>
>>>
>>>> Hi James,
>>>>
>>>> Thanks for this explanation.
>>>>
>>>> You write "they are different snapshots of a common, underlying thing". What is that common, undelying
>>>> thing? That's where I struggle.
>>>>
>>>> Luc
>>>>
>>>>
>>>> On 14/11/2012 05:44, James Cheney wrote:
>>>>
>>>>> I agree with Simon that the text and response was approved (and there was plenty of time for people to check it).  However, I also recall wondering if the file example was correct or misleading.
>>>>>
>>>>> I think it's a matter of perspective: if we consider a file that is "moved" from /a/b at time 1 to /c/d at time 2, then asserting that they are alternates is a way of saying that they are different snapshots of a common, underlying thing.  This is valid:
>>>>>
>>>>> entity(f1,[time=1,loc=/a/b,content="foo"])
>>>>> entity(f2,[time=2,loc=/c/d,content="foo"])
>>>>> wasDerivedFrom(f2,f1,[op=move])
>>>>> alternateOf(f1,f2)
>>>>>
>>>>> However, if we consider a file that is "copied" from /a/b at time 1 to /c/d at time 2, with the original at /a/b being updated  at time 2, then it may be confusing to say this:
>>>>>
>>>>> entity(f1,[time=1,loc=/a/b, content="foo"])
>>>>> entity(f1',[time=1,loc=/a/b, content="bar"])
>>>>> entity(f2,[time=2,loc=/c/d, content="foo"])
>>>>> wasDerivedFrom(f1',f1,[op=update])
>>>>> wasDerivedFrom(f2,f1,[op=copy])
>>>>> alternateOf(f1,f2) // maybe?
>>>>> alternateOf(f1,f1')  // maybe?
>>>>>
>>>>> However, this isn't invalid (with any combination of the last two assertions).  This is because adding alternate assertions doesn't lead to any other new inferences - we don't have a way of telling whether two entities overlap in time or have inconsistent attributes.
>>>>>
>>>>> If we say that they are all alternates, this amounts to Simon's view - the location being a transient attribute of the totality of all versions of the file anywhere.  If we only assert one or the other then we're taking a narrower view of the file as a linear sequence.  Although the semantics I had in mind originally assumes that alternate entities are along a linear "timeline", there is nothing in the constraints that forces this.
>>>>>
>>>>> Maybe instead of replacing the file example, we could present the two different takes on it?
>>>>>
>>>>> --James
>>>>>
>>>>> On Nov 13, 2012, at 9:27 PM, Luc Moreau wrote:
>>>>>
>>>>>
>>>>>
>>>>>> No problem with not using content negotiation.
>>>>>>
>>>>>> But the question is: is the current example correct? I have a doubt.
>>>>>>
>>>>>> Luc
>>>>>>
>>>>>>
>>>>>> On 13/11/2012 15:09, Ivan Herman wrote:
>>>>>>
>>>>>>
>>>>>>> (also echoing the concerns some others had...)
>>>>>>>
>>>>>>> Question: What is the percentage of the expected audience of this primer who really understand what content negotiation is all about? My answer is: not very high:-(
>>>>>>>
>>>>>>> Seriously, content negotiations and mime time may be a black art for many who would want to use our ontology.
>>>>>>>
>>>>>>> (I may be wrong, I did not make some sort of a scientific study...)
>>>>>>>
>>>>>>> Ivan
>>>>>>>
>>>>>>> On Nov 13, 2012, at 09:46 , Luc Moreau wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> When preparing the PROV tutorial at ISCW'12, we discussed the example for alternate in the primer.
>>>>>>>>
>>>>>>>>    , if a file is copied from one directory to another, we may want to say that (according to our model) these are both the same file, just in a different location. We may say that the file in the first directory, F1, is an alternate of the file in the second directory, F2. Note that it is the context (location) rather than content of the file that differs between the entities in this case.
>>>>>>>>
>>>>>>>> We didn't find this example as the most compelling.
>>>>>>>>
>>>>>>>>   I don't think that in general, copying a file creates  an alternate. If file f2 is a copy of file f1, I don't know what same thing, f1 and f2 present a same aspect of. And while there is some form of caveat "accordign to our model", this example may be confusing for readers.
>>>>>>>>
>>>>>>>> Instead of this example, we decided to use a content negotiation example:
>>>>>>>>
>>>>>>>> Dereferencing a url requesting different mime types eg. turtle or rdf/xml, returns two entities that are alternate of each other.
>>>>>>>>
>>>>>>>> If we are in agreement, can we change the primer accordingly?
>>>>>>>>
>>>>>>>> thanks,
>>>>>>>> Luc
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> 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
>>>>>>>>
>>>>>>>> United Kingdom
>>>>>>>> http://www.ecs.soton.ac.uk/~lavm
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>> ----
>>>>>>> Ivan Herman, W3C Semantic Web Activity Lead
>>>>>>> Home: http://www.w3.org/People/Ivan/
>>>>>>> mobile: +31-641044153
>>>>>>> FOAF: http://www.ivan-herman.net/foaf.rdf
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> --
>>>>>> 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
>>>>>> 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
>>>> 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
>> 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
United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
Received on Monday, 19 November 2012 10:48:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 19 November 2012 10:48:18 GMT