RE: PROV-ISSUE-337 (agent-and-entity): agent should not be a subclass of entity [prov-dm]

Hi Luc,

OK. Personally, I find that a little odd...

Is it that we're saying (a) someone might not want to bother describing whether some agent is an entity or an activity, or (b) that there is actually stuff that is neither an entity or an activity but can have had responsibility? 

I think (b) could lead to difficulties. If we imply that there are past occurrences that are neither entities nor activities, then does PROV allow us to say anything about these except that they had responsibility? It would seem a little mysterious, and I'm not clear how we would justify it.

(a) is more comprehensible, but if that's what we mean, I suggest we be explicit, e.g. "An agent is either an entity or an activity, but it may not be described which for any given agent." Even then, I don't know why we wouldn't require its type to be explicit - the PROV creator must know, it seems little to ask, and it could greatly aid interoperability as otherwise others have to guess what seems quite fundamental information for the provenance graph.

thanks,
Simon

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

Requirements for Provenance on the Web:
http://eprints.dcs.kcl.ac.uk/1399/
________________________________________
From: Luc Moreau [L.Moreau@ecs.soton.ac.uk]
Sent: 03 April 2012 13:43
To: public-prov-wg@w3.org
Subject: Re: PROV-ISSUE-337 (agent-and-entity): agent should not be a   subclass  of entity [prov-dm]

Hi Simon,

My proposed definition may not have been clear, but yes.

Luc



On 04/03/2012 09:32 AM, Miles, Simon wrote:
> ...a follow-up question:
>
> Can something be an agent but neither an entity or an activity?
>
> thanks,
> Simon
>
> Dr Simon Miles
> Senior Lecturer, Department of Informatics
> Kings College London, WC2R 2LS, UK
> +44 (0)20 7848 1166
>
> Requirements for Provenance on the Web:
> http://eprints.dcs.kcl.ac.uk/1399/
> ________________________________________
> From: Miles, Simon [simon.miles@kcl.ac.uk]
> Sent: 03 April 2012 09:21
> To: Provenance Working Group
> Subject: RE: PROV-ISSUE-337 (agent-and-entity): agent should not be a  subclass of entity [prov-dm]
>
> Hello Luc,
>
> I agree with agency (here, responsibility) being independent of the nature of the agent. I also like the suggested definition.
>
> In particular, I'd say the WG does not want to tie itself to one side of philosophical debates regarding responsibility - it is just not something we should be concerning ourselves with in a provenance standard. By allowing responsibility to be ascribed to individuals and not to actions, we are taking (and having to defend) a stance that is not particular to provenance. If Bob ran away because Alice shouted, then was "Alice" responsible or was "Alice shouting" responsible, or both? Was the First World War responsible for the death of millions, or was it the individuals who instigating the battles, or both? Or can we, in PROV, avoid deciding :-)
>
> For the sake of explanation, I think we could find a better example of a responsible activity than collaboration. An act of collaboration is an activity, but a collaboration itself could be considered an entity.
>
> thanks,
> Simon
>
> Dr Simon Miles
> Senior Lecturer, Department of Informatics
> Kings College London, WC2R 2LS, UK
> +44 (0)20 7848 1166
>
> Requirements for Provenance on the Web:
> http://eprints.dcs.kcl.ac.uk/1399/
> ________________________________________
> From: Provenance Working Group Issue Tracker [sysbot+tracker@w3.org]
> Sent: 02 April 2012 22:53
> To: public-prov-wg@w3.org
> Subject: PROV-ISSUE-337 (agent-and-entity): agent should not be a subclass of entity [prov-dm]
>
> PROV-ISSUE-337 (agent-and-entity): agent should not be a subclass of entity [prov-dm]
>
> http://www.w3.org/2011/prov/track/issues/337
>
> Raised by: Luc Moreau
> On product: prov-dm
>
> Currently, prov-dm defines agent as follows:
>
> An agent is a type of entity that bears some form of responsibility for an activity taking place. An agent is a particular type of Entity. This means that the model can be used to express provenance of the agents themselves.
>
> While it is nice to be able to express the provenance of agents, it is not obvious to me that agents should always be entities.  In fact, they could be activities.
>
> Consider a collaboration activity, to which several agents ag1, ag2, ..., agn are associated. Why can't we see it as an agent too?
> activity(collaboration)
> wasAssociatedWith(collaboration,agi,contract)
>
> agent(collaboration)
> wasAttributed(nice-piece-of-work,collaboration)
>
>
> So, I would propose the following alternative definition:
>
>
> An agent is something that bears some form of responsibility for an activity taking place.
>
> A given agent may be a particular type of Entity or Activity. This means that the model can be used to express provenance of the agents themselves.
>
>
> Looking at prov-o, I notice that they have already defined an agent as subclass of owl:Thing.
>

--
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 Tuesday, 3 April 2012 13:08:53 UTC