- From: Dave Reynolds <dave.e.reynolds@gmail.com>
- Date: Mon, 22 Oct 2012 15:19:37 +0100
- To: public-gld-wg@w3.org
On 22/10/12 13:26, Joćo Paulo Andrade Almeida wrote:
> Dear All,
>
> I am currently joining the GLD working group and looking forward to
> contribute.
>
> I have a special interest in the Org Ontology as we have been working on
> the modelling of organizational structures, roles and the concept of
> organization from the perspective of foundational ontologies. (See refs
> below.)
Interesting. Org has just been voted to Last Call.
I don't know what the process implications are here!
> The first thing that caught my attention was the "reportTo" relation in
> the Org ontology.
>
> Currently, the domain and range are a union of foaf:Agent and org:Post.
> (Where in the previous April 2012 version it was only foaf:Agent).
>
> In my opinion, foaf:Agent should be excluded from the domain and range,
> because "reportsTo" is always a relation in the scope of organizations and
> never a relation of Agents outside the context of membership in an
> organization. Let me clarify this with an example: Alex may report to Joćo
> Paulo in the scope of his membership to UFES, but Joćo Paulo may report to
> Alex in the scope of his membership to The Victoria Chess Club.
Firstly, one reason for going for the union rather than simply
restricting the range to org:Post was for backward compatibility. This
is, after all, a ontology that has been in use for a couple of years,
not something that was developed by this working group. So some care has
to taken over changing things.
Secondly, while I agree that the more useful relation is specific to the
organizational context (hence org:Post) that doesn't mean that the
simpler usage should be necessarily excluded.
As always with ontologies there is a trade-off between completeness of
representation and simplicity for broad adoption.
You could equally say that "reportsTo" is always scoped to a time
period. That is equally true but the current modelling does not provide
time qualified reporting relationships.
Furthermore I have personally had multiple reporting relationships
concurrently within the same organization with differences between
"dotted" and "direct" reporting, differences between "technical" and
"line" reporting, and differences in reporting between the different
organization roles that I concurrently held. Simply qualifying a
reporting relationship to an organization does not cover all of that
complexity.
In my view a really nuanced representation of reporting relationships
should be an extension ("profile") to ORG and is out of scope for the
current work.
> So the domain and range of "reportsTo" should be foaf:Membership or
> foaf:Post (or the union).
I assume you meant org:Membership and org:Post.
Certainly can't be org:Membership, that is an essentially a tuple in an
n-ary relation and not really something an agent can report to :)
Dave
Received on Monday, 22 October 2012 14:20:09 UTC