Re: Problems with Example 2 and 4 in https://www.w3.org/TR/2013/REC-prov-o-20130430

Adding to the already exhaustive replies below, one reason why Entities and Agents are not disjoint classes is to make it possible to express the “provenance of agents” within the same context where the agents operate.
For example, agents may have a “state” that changes following some Activities they perform, and they are responsible for other Activities while in a certain state. Let’s say yesterday I knew nothing about provenance but I then educated myself (through a provenanceReadUp Activity) which Used the PROV documents and generated a new version of myself, i.e. a new _provenanceExpert Entity, which is actually the Agent who today is responsible for this very Activity, namely replying to this thread and thus generating a new Entity representing this email.
So if you ask the question: “if I look at the provenance of this email I find that it includes a response by some Paolo Agent, but what entitles this guy to write such a convoluted self-referential response on the subject of provenance? And you could consult the provenance of me today and discover I actually did do my homework (just in time to reply to this).

I hope I have thought it through – happy to stand corrected by the real experts!

--Paolo

Ps @Stian Soiland-Reyes<mailto:soiland-reyes@manchester.ac.uk>:
Rather any one particular Agent - in a particular PROV bundle - that is
either implicitly (through other PROV relations) or explicitly (type)
also an Actity must then not also be implicitly or explictly an
Activity, or vice versa.

Did you mean Actity --> Entity?


From: Stian Soiland-Reyes <soiland-reyes@manchester.ac.uk>
Date: Tuesday, 3 August 2021 at 13:11
To: "nicolas.f.rouquette@jpl.nasa.gov" <nicolas.f.rouquette@jpl.nasa.gov>, "jcheney@inf.ed.ac.uk" <jcheney@inf.ed.ac.uk>
Cc: "j.s.jenkins@jpl.nasa.gov" <j.s.jenkins@jpl.nasa.gov>, W3C Prov <public-prov-comments@w3.org>, Luc Moreau <luc.moreau@kcl.ac.uk>, "david.a.wagner@jpl.nasa.gov" <david.a.wagner@jpl.nasa.gov>
Subject: Re: Problems with Example 2 and 4 in https://www.w3.org/TR/2013/REC-prov-o-20130430

Resent from: W3C Prov <public-prov-comments@w3.org>
Resent date: Tuesday, 3 August 2021 at 13:10

⚠ External sender. Take care when opening links or attachments. Do not provide your login details.

James is right on spot, as always.

You can't conclude the generic disjointness of Agent being
Entity/Activity on a class level or even at instance level.

Rather any one particular Agent - in a particular PROV bundle - that is
either implicitly (through other PROV relations) or explicitly (type)
also an Actity must then not also be implicitly or explictly an
Activity, or vice versa.

For instance a software agent can either be considered an Entity
(someting you download, a file, a chunk of RAM), or an Activity
(something that was started and stopped, that consumes and generates
entities).

This is basically a distinction between something's physical footmark -
e.g. me as a set of atoms; vs what is being done - e.g. my
personality/thoughts - or even just my "processing emails" activity in
case you identified me by my shorter-lived email address.

It is fine to leave this distinction undeclared, as in the open world
assumption it just means the modeller has not taken a stand for that
instance.

That does not mean there always must be a type of activity or entity
for any given agent. Perhaps multiple activities and entities are
needed to explain a particular agent.

However it means that when combining or validating PROV bundles you
need to take care not to cause an agent to simultanuosly be both an
activity and entity - that could give the agent two alternate and
probably quite different provenance traces (e.g. the software run of my
email client was started today by me, while the code was generated 7
April by an Ubuntu maintainer)


On Tue, 2021-08-03 at 10:39 +0100, James Cheney wrote:
Dear Nicolas,

I'm sure others on this list will have more apposite responses, but
as far as I am aware it was an intentional design decision NOT to
make entity and agent disjoint (and similarly, NOT to make activity
and agent disjoint).  Only entity and activity are disjoint.  The
paper "The rationale of PROV" [1] explains that the group decided to
minimize the number of disjointness constraints (requirement GE1),
and later explains that activities and entities are disjoint
(requirement VI7) but agents are allowed to be entities or activities
(requirements VI3, VI4).  I am not sure whether this decision is
explained explicitly in the PROV-O recommendation itself, but you can
also see this recorded, for example, in the (non-normative) semantics
document [2]:

> 3.2.3 Agents
> An agent is an object that can act, by controlling, starting,
> ending, or participating in activities. An agent is something that
> bears some form of responsibility for an activity taking place, for
> the existence of an entity, or for another agent's activity. Agents
> can act on behalf of other agents. An agent may be a particular
> type of entity or activity; an agent cannot be both entity and
> activity because the sets of entities and activities are disjoint.
>

Thus, a heuristic that assumes that classes that have no common
subclasses ought to have been declared disjoint, would not provide
good advice in this case.

--James

[1] Luc Moreau, Paul Groth, James Cheney, Timothy Lebo, Simon Miles:
The rationale of PROV. J. Web Semant. 35: 235-257 (2015)
[2] https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.w3.org%2FTR%2Fprov-sem%2F&amp;data=04%7C01%7Cpaolo.missier%40newcastle.ac.uk%7C6d6aa8173dd8474c4bfe08d95677d667%7C9c5012c9b61644c2a91766814fbe3e87%7C1%7C0%7C637635894961008207%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=WwO8MSXJDRi4btbKY%2BCsbMeBMSiibWRjBAwlZtLzXhw%3D&amp;reserved=0


> On Aug 2, 2021, at 11:23 PM, Rouquette, Nicolas F (US 319K) <
> nicolas.f.rouquette@jpl.nasa.gov<mailto:nicolas.f.rouquette@jpl.nasa.gov>> wrote:
>
> This email was sent to you by someone outside the University.
> You should only click on links or attachments if you are certain
> that the email is genuine and the content is safe.
> I have encoded the complete Prov-O ontology from the specification
> (https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.w3.org%2FTR%2F2013%2FREC-prov-o-20130430&amp;data=04%7C01%7Cpaolo.missier%40newcastle.ac.uk%7C6d6aa8173dd8474c4bfe08d95677d667%7C9c5012c9b61644c2a91766814fbe3e87%7C1%7C0%7C637635894961008207%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=C5AdmcHLiD%2FDWnngIFTOJ9muNocLDeQhzdkzLSsqLcg%3D&amp;reserved=0) using our
> Ontological Modeling Language, OML (https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.opencaesar.io%2Foml%2F&amp;data=04%7C01%7Cpaolo.missier%40newcastle.ac.uk%7C6d6aa8173dd8474c4bfe08d95677d667%7C9c5012c9b61644c2a91766814fbe3e87%7C1%7C0%7C637635894961008207%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=UMIrJ%2FgalB5U6Bs5KZ6pWrgbWj2T3irTU0lYHPgLnlE%3D&amp;reserved=0)
> that we created to facilitate ontology authoring for humans by
> automating the process of calculating disjointness axioms according
> to a simple policy: if two classes have no common specialization in
> the scope of a set of ontologies, then they are asserted to be
> disjoint in that scope. Applying this policy to an OML vocabulary
> of Prov-O helped identify two problems:
> In Example 2:
> https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fopencaesar%2Fprovenance-vocabularies%23example-2&amp;data=04%7C01%7Cpaolo.missier%40newcastle.ac.uk%7C6d6aa8173dd8474c4bfe08d95677d667%7C9c5012c9b61644c2a91766814fbe3e87%7C1%7C0%7C637635894961008207%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=rCHbjoISwkNp7sbug5iM8AwoZAWl5k%2BLETt%2BU5Fr1gY%3D&amp;reserved=0

> In Example 4:
> https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fopencaesar%2Fprovenance-vocabularies%23example-4&amp;data=04%7C01%7Cpaolo.missier%40newcastle.ac.uk%7C6d6aa8173dd8474c4bfe08d95677d667%7C9c5012c9b61644c2a91766814fbe3e87%7C1%7C0%7C637635894961008207%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=blbC1ZB%2BG4YkwJM6okNBeEzKcoy16TbWMuEpRVSLZso%3D&amp;reserved=0

>
> I understand that the working group that developed the provenance
> ontology has long been dismantled; however, I believe that it would
> be helpful to fix these problems as the examples convey a subtly
> incorrect idea about the domain/ranges of the properties involved.
>
> Nicolas F. Rouquette
> Principal Computer Scientist
> Jet Propulsion Laboratory, California Institute of Technology M/S
> 301-490, 4800 Oak Grove Dr, Pasadena, CA 91189, USA
> email: nicolas.f.rouquette@jpl.nasa.gov<mailto:nicolas.f.rouquette@jpl.nasa.gov>
> phone: +1 (818) 354-9600
> cell: +1 (626) 639-5282

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.