Re: SUO: Organization
Ian --
Your proposed changes seem to answer some of the
questions raise, but there is one clarification of the
meaning of "Agent" that I think is needed. I think
there is an important distinction that needs to be
made between physical entities that *can* show
agency, the *predicate* of agency, and the designation
of a specific agent as the agent of a specific action.
The CYC #$Agent starts with the statement that an
#$Agent is something that *can* show independent action;
but the Docs then exclude plants.
In SUMO there is a class "Agent" and a predicate
"agent" which relates the role to the class. The
question is how to interpret the class "Agent".
The documentations says, like the CYC:
"An agent is something or someone that can
act on its own and produce changes in the world"
And the axiom indicates that there has to be some
action that an agent is agent of:
(<=>
(instance-of ?X Agent)
(exists (?Y)
(agent ?Y ?X)))
This still leaves open the interpretation
(which I prefer) that an Agent does not have
to be an agent *at every point in time*,
merely to *have been* in the role of agent at some
time, with a potential to be so again. I think
that is consistent with the axiom above.
If this interpretation is what is intended, then
every life form should be an agent, and every
group of living things should also be an agent,
because they are capable of agency -- though they
may not exhibit obvious agency at every point
in time. At the very least, they are agents of
the metabolism of nutrients, and of biochemical
processes; in fact, for every living thing
there is probably some biochemical process at
every point in time of which it is an agent.
Thus, I think, every Group should still be an Agent,
though the GroupAgentFn could still be useful to
generate Agents from other collections, which
might not necessarily be Agents. (this assumes
that non-living things can be Agents. We discussed
before such a general type of Agent.)
On the question of the relations of an Organization
to its members, I am trying to put together a short
comment relevant to that, in response to Chris
Partridge's note.
Pat Cassidy
===============================
Ian Niles wrote:
>
> Hi All,
>
> I wanted to summarize the various criteria that people have proposed for any
> adequate formalization of the concept of organization and then sketch a
> formalization that satisfies these criteria.
>
> Here, then, are the various criteria that an organization satisfies.
>
> 1. An organization has agency, e.g. it exhibits intentionality, and
> it has rights, responsibilities, and obligations.
>
> 2. An organization may have members, but it is not required to have
> members. There are many examples of organizations, e.g. corporations and
> churches, that have assets are liable for certain claims etc, even though
> they have no members.
>
> 3. An organization has temporal extent. It comes into being at a
> certain point in time, and it goes out of existence at another
> point.
>
> 4. An organization can have various sorts of members. Owners are
> members of organizations, and employees, directors, and other
> stakeholders may also be members of organizations (although perhaps in
> different senses). An organization may also have other organizations as
> members.
>
> All of these criteria have been extracted from Pat Cassidy's and Chris
> Partridge's emails on the subject. The current SUMO formalization of
> 'Organization' as a subclass of 'Collection' satisfies points 1 and 3,
> because 'Organization' is a subclass of 'Group' and 'Group' is a subclass of
> 'Agent' (point 1) and because 'Organization' is indirectly a subclass of
> 'Physical' (point 3). However, since 'Organization' is a subclass of
> 'Collection', it is required to have members, so point 2 is not satisfied
> completely. Furthermore, there is no provision for the different sorts of
> members that may make up an 'Organization', so point 4 is also apparently
> not satisfied.
>
> My new proposal is to cleanly separate the notions of agency and group.
> Rather than make 'Organization' a subclass of 'Group', I think we should
> make it a direct subclass of 'Agent', where it will be understood as a class
> of legal entities with certain rights, responsibilities, intentionality,
> etc. We should also, I think, remove the subclass link between 'Group' and
> 'Agent', because there are groups of animals and humans that do not exhibit
> agency of any sort, e.g. a group of cows grazing. We can then use a new
> function, defined as follows, to relate 'Groups' to 'Agents'.
>
> (instance GroupAgentFn UnaryFunction)
> (domain GroupAgentFn 1 Group)
> (range GroupAgentFn Agent)
> (documentation GroupAgentFn "Assigns an instance of 'Agent' to an instance
> of 'Group'. In some cases, the 'Agent' assigned will be identical to the
> group, e.g. a flock of geese flying northward. In some cases, the 'Agent'
> will be different from the 'Group', e.g. the 'GroupOfPeople' making up an
> 'Organization' is distinct from the legal entity that is the 'Agent'. Note
> that this is a partial function. There are many cases of 'Groups' which do
> not exhibit agency.")
>
> I think this proposal addresses points 1, 2, and 3 above, and I think it has
> the advantage of making the overall structure of the ontology cleaner and
> clearer. It is important to note, however, that this proposal does nothing
> in the way of answering point 4. We are still stuck with the problem of
> defining various sorts of relations between organizations and the different
> classes of members that make them up. However, this problem seems to me to
> be separable from the problem of figuring out what an "organization" is. In
> fact, as I see it, we can take it as the problem of defining subrelations of
> 'member', since 'member' encompasses all of the relations that we would ever
> want to define between an organization and a person who belongs to the
> organization. The only exception to this, as I see it, is the
> 'subOrganizations' relation. In this case, the relation could be redefined
> so that it is no longer a subrelation of 'subCollection'.
>
> I hope this proposal makes sense. Let me know if and where you disagree
> with it.
>
> -Ian
>
>
--
=============================================
Patrick Cassidy
MICRA, Inc. || (908) 561-3416
735 Belvidere Ave. || (908) 668-5252 (if no answer)
Plainfield, NJ 07062-2054 || (908) 668-5904 (fax)
internet: cassidy@micra.com
=============================================