W3C home > Mailing lists > Public > public-prov-wg@w3.org > June 2011

Re: PROV-ISSUE-19: is this observable or not observable?

From: Simon Miles <simon.miles@kcl.ac.uk>
Date: Thu, 9 Jun 2011 12:01:29 +0100
Message-ID: <BANLkTi=ZVpOC7uQz7n1QHah9ntGhFjA4Lg@mail.gmail.com>
To: Provenance Working Group WG <public-prov-wg@w3.org>
Luc,

+1 (may or may not be observable), though I must admit I'm finding
this thread a bit ungrounded, so I'm not sure of the consequences.

I would note that it is surely critical that we can model cases of
knowing provenance without knowing what it is the provenance of. For
example, the provenance of a hospital patient may include their
treatments and appointments. The patient's details may be anonymised,
and to check privacy controls, we should be able to ask whether you
can identify the patient from their provenance alone. This means
putting ourselves in the shoes of someone observing the provenance but
not the patient.

I think I agree with Graham's last argument regarding the provenance
of conceptual information and suggest that "authoritative information"
is just another way of saying "invariant property" (i.e. what makes an
IPVT invariant). I'd argue that you have to know the invariant
properties of the IPVT to honestly assert its provenance, implying
observability in some sense (direct/indirect), but this does not apply
to someone querying or annotating the provenance, so should not be a
restriction in the model.

Hope that makes sense...

Thanks,
Simon

On 8 June 2011 22:04, Luc Moreau <L.Moreau@ecs.soton.ac.uk> wrote:
>
> Hi Graham,
>
> I personally had never considered observability as essential property in
> the context of provenance.
> I am fine with the option that you propose.  I would even go further,
> and venture the following proposal.
>
> PROPOSED: IVPT may or may not be observable
>
> What do you think?
>
> Regards,
> Luc
>
>
> On 08/06/11 18:31, Graham Klyne wrote:
>> Hi Luc,
>>
>> Luc Moreau wrote:
>> > The reason why I raised the issue is that over the WE, when discussing
>> > with Kai, this notion of observability popped up.
>>
>> OK.
>>
>> My take on "observability" would be along the lines of there being an
>> available method by means of which information directly about the
>> thing observed could be obtained.
>>
>> For concepts, I would think this amounts to some way in which
>> authoritative information about the concept can be accessed.  For
>> concepts for which there is no such authoritative information, then
>> they aren't (directly) observable, but may be "indirectly observable"
>> (which I would treat as not observable) though (e.g.) assertions made
>> by other people.
>>
>> Example: "truth".  I don't suppose there's any way in which this
>> concept is directly observable.  But there are any number of
>> philosophical discourses on truth that might be cited as giving us
>> indirect information about truth (e.g. Tarski, Quine, etc.)
>>
>> This suggests to me that we may want to have an identifier for the
>> concept, even if there are no direct "observations" associated with
>> it;  I hazard that it has no direct provenance.  But we can still say
>> that Tarski, Quine, etc. say things about the concept of truth.
>>
>> It's a viewpoint ... with which you may reasonably disagree.  But I'd
>> hate us to get hopelessly tangled in this debate when there are other
>> useful things we can make progress on.  Would it be an option to say
>> that for some things we don't (yet) know whether or how they can be
>> observed?
>>
>> #g
>> --
>>
>>
>> Luc Moreau wrote:
>>>
>>> Hi Graham,
>>>
>>> The reason why I raised the issue is that over the WE, when
>>> discussing with Kai, this
>>> notion of observability popped up. I think Jim also mentioned it in
>>> another thread (apologies,
>>> if I got it wrong).  In all fairness, I thought we had to discuss this.
>>>
>>> Given that we have indicated that we want to track the provenance of
>>> things, which may
>>> be physical, digital, CONCEPTUAL or otherwise, I don't know what
>>> observability means
>>> when things are conceptual.
>>>
>>> I take note of Carl's pointers to definitions of observability in the
>>> physical world.
>>>
>>> I would argue that even in the digital world, observability is not
>>> straightforward.  In the provenance
>>> challenge, we have seen techniques instrumenting code, i.e. adding
>>> constructs to record provenance.
>>> In that case, can we say the system observed what was happening? or
>>> was it programmed to
>>> record provenance synchronously with its execution?
>>>
>>> Regards,
>>> Luc
>>>
>>>
>>> On 07/06/11 12:06, Graham Klyne wrote:
>>>> May I suggest we see if this is an issue in light of the proposed
>>>> definitions?
>>>>
>>>> #g
>>>> --
>>>>
>>>> Luc Moreau wrote:
>>>>>
>>>>> Dear all,
>>>>>
>>>>> When we discussed the notion of 'Invariant View or Perspective on a
>>>>> Thing, there were
>>>>> suggestions that it should be observable, and counter-suggestions
>>>>> that it should not be.
>>>>>
>>>>> It would be good to discuss both sides of the argument, in an
>>>>> attempt to reach consensus.
>>>>>
>>>>> Best regards,
>>>>> Luc
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>
>
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________
>



-- 
Dr Simon Miles
Lecturer, Department of Informatics
Kings College London, WC2R 2LS, UK
+44 (0)20 7848 1166
Received on Thursday, 9 June 2011 11:01:57 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 13:06:31 GMT