Dear Simon,
Thank you for your proposal! Just a few comments:
Please mark a proposal meant to be formally decided as "ISSUE" in the
subject.
Since your proposal is a nice example of an AI / Semantic Web school of
thought, let me
elaborate here the quite fundamental difference of the CRM approach,
which over the past decade
has been verified by practice to be much more robust for real life
information integration.
Frequently the former school of thought takes our approach as lack of
formality. In reality, our arguments
are much more complex.
Indeed the identity of E74 Group is not at all determined by its
members, at least not in the current
definition.
In the CRM methodology (still to be published...:-[ ), we should
distinguish between:
a) substance : the Group is "constituted" or "consists of" persons, like
a material. The mathematical
phrase : "Let a <group> be an aggregation" is already a
classification to my understanding.
I do not really see a point why we should define the term <group>
just as a synonym of aggregation.
Isn't the term "aggregation" good enough?
If the additional identity condition is not known ("not determined
solely"), then in an open World
this does not add anything verifiable to the term "aggregation",
and hence does not define anything.
b) identity conditions: the Group is identified by the intention or fact
of collective behavior. As long
as there is a continuity in this, it exists. In this sense,
the office of a president may be vacant
for some time, but the next president will speak again as
"The President".
In contrast, if I take a picture from a market place, the people on
the photograph do not form
a Group by the fact that they appeared by chance in the focus of my
lens. They have nothing in
common due to this fact an independent observer could confirm.
Therefore, such definition of
a Group is irrelevant for information integration. (Note the
functional restrictions of the CRM!)
c) begin/end of existence criteria: A Group may be formed (come into
existence) by explicit decision or spontaneous collective acctivity.
d) Unity: parts belong to the group as long as they commit to the
collective activity or are formally declared as parts.
(My wordings above are of course open to discussion...).
If it wouldn't be a gigantic task, we should rewrite all scope notes to
distinguish these 4 facets of definition.
Please let me repeat, that in the CRM "This restriction is not reflected
in the label" is not an argument.
We do have to require that people read the scope notes. There is
absolutely no way to pack definitions into
labels, so we have abandoned this attempt. Our problem is didactic here.
How to make people understand not to seek definition in labels? We even
discussed deleting the labels all over. We could equally talk about
"Collective Behavior Actor Group" etc. until the whole scope not is in
it, or better just "E74".
We do not have other "groups" in the CRM, and the CRM does not intend to
interpret language, but intends to
provide a controlled language for describing data structures made by
scholars and IT people for integration.
The possible meanings of a term like <group> is explicitly out of scope
of the CRM.
Finally, please do not propose "camel case" labels, scholars do not
appreciate their elegance....;-)
Comments welcome,
Martin
On 28/4/2014 10:21 μμ, Simon Spero wrote:
*Title* *Rename E74 from Group to GroupOfActors*
*Background* Let a <group> be an aggregation of one or more members,
where the identity of the <group> is not determined solely by the
members of the <group> (thus distinguishing <group> from <set>).
From the definition of P107, it is clear that every E74 refers to
<group>; otherwise the existence of a former member would imply that
the group no longer exists.
However, E74 has stronger semantics beyond that of a basic <group>.
Every member of an instance of E74 is a <group> all of whose members
are E39_Actor's.
This restriction is not reflected in the label of E74, and can lead
to confusion.
*Old Proposal*
*Current Proposal* Classes whose instances are <group>, all of whose
members are instances of a specified class <MemberClass> should be
given labels of the form "Group of <MemberLabel-pl>" where
<MemberLabel-pl> is the plural form of the label of <MemberClass>.
For E74, MemberClass is Actor; this proposal would thus assign E74
the label "Group Of Actors".
If OWL is used, a single membership property can be used, with
restriction classes used to specify member type.
(this also suggests that P107.1 may be awkwardly named, since it
appears to denote the role of a member within a group (a members role
in a group may change over time)
*Outcome*
*Status* proposed
*Working Group*
*Starting Date*
*Closing Date*
[There are other changes I would suggest making to E74 separate from
this proposal.
- allowing groups to consist of only a single member
- considering whether a group may become dormant (have zero members),
then gain new members, yet retain its identity as the same group -
that is, whether the group can be temporally discontinuous. This would
cover, for example, a student society that publishes journals, , is
inactive for a year or so, then reforms.
There are also issues with P107 and P107.1 ]
Simon
On Mon, Apr 28, 2014 at 11:31 AM, martin <[email protected]
<mailto:[email protected]>> wrote:
Dear Joao,
Your comments well taken! - one of the most important principles
of the CRM is that
the label is never a definition. In 15 years, we could never
reconcile linguistic values with
clear semantics under the functional restrictions of the CRM.
The label is only a mnemonic, the definition is exlusively the
scope note, and the identity
exclusively the E-number. If this could be better pointed out in
our documentation,
comments are MOST welcome.
The definition of "group" by Guarino& Welty has nothing to do with
E74 Group. Actually,
he is talking about what other ontologies would call an
"aggregate", like ORE's
"ore:aggregation", or an enumeration, sometimes even a collection.
We cannot avoid that other ontologies use different senses for the
same word. However, we commit to the point Guarino is making here:
E74 Group is "constituted by" an aggregate of persons. The CRM
property "has former or current member" is a kind of constitution
exactly in the sense of Guarino.
Our reasons to use the label "Group" was to be close to one
natural use of the term, such as
a "Working Group", a "group" of (joined) industries, a group of
hikers, a discussion group etc. which imply common intentions,
temporarily or permanently.
It's always a hard decision to coin new terms, which make the
ontology appear more and more alien, but do not relieve from good
definitions in the end.
Any renaming proposals are also always welcome, and will be
treated as issues.
Best wishes,
Martin
On 28/4/2014 4:15 μμ, João Oliveira Lima wrote:
Dear Stephen Stead and Simon Spero,
Thank you for your response.
The picture is clearer but some doubts remains.
Maybe the term "Group" is not best to denominate the "Collective
Actor" because the term "Group" is intrinsically tied with the
"constitution" idea. See, for example, the follow example
extracted from "An Overview of Ontoclean" (Guarino & Welty,
Handbook on Ontologies, Springer Verlag, (2004)):
"Take for instance two typical examples of social entities, such
as a bridge club and a poker club. These are clearly two separate
entities, even though precisely the same people may participate
in both. Thus we would have a state of affairs where, if the
social entity was the group of people, the two clubs would be the
same under the identity criteria of the group, and different
under the identity criteria of the social entity. Note also that
if a club changes its members it is still the same club, but a
different group of people. The solution to the puzzle is that
this is, once again, a constitution relationship: a club is
constituted of a group of people.".
In addition, it’s possible to talk about an instance of "E40
Legal Body" that was constituted by only one "E21 Person".
I've seen now that the FRBRoo
'F15 Complex Work' R10_has_member 'F1 Work'
has similar membership structure:
'E74 Group' P107_has_current_or_former_member 'E39 Actor'.
Joao Lima
On Sun, Apr 27, 2014 at 10:50 PM, Stephen Stead
<[email protected] <mailto:[email protected]>> wrote:
Joao Lima
Do not be fooled by the name E74 Group! E74 Group is by
definition groups of actors. The group of legislation (each
an E28 Conceptual Objects)that you are mention is itself,
another E28 Conceptual Object). The part decomposition
function (using the appropriate Properties depending on the
sort of things that have whole-part relationships) generally
deals with this kind of thing.
Hope This Helps
SdS
Stephen Stead
Tel +44 20 8668 3075 <tel:%2B44%2020%208668%203075>
Mob +44 7802 755 013 <tel:%2B44%207802%20755%20013>
E-mail [email protected] <mailto:[email protected]>
LinkedIn Profile http://uk.linkedin.com/in/steads
*From:*Crm-sig [mailto:[email protected]
<mailto:[email protected]>] *On Behalf Of *João
Oliveira Lima
*Sent:* 28 April 2014 01:04
*To:* crm-sig
*Subject:* [Crm-sig] E74 Group - Generalization
I was wondering if the class "E74 Group" could be generalized
as follows:
"E74 Group"
Subclass of "E1 Entity";
"P107 has current or former member (is current or former
member of)"
Domain "E74 Group"
Range "E1 Entity".
With this generalization would be possible to represent
groups of any entities (not just E39 Actors).
For example, in the legislative process, a bill may be part
of a group of bills that move together, as they dealt with
similar matters. The bill group membership (or the exclusion)
is formalized by a document (petition). In the field of
cultural heritage, there are examples of groups like this?
Regards,
Joao Lima
_______________________________________________
Crm-sig mailing list
[email protected] <mailto:[email protected]>
http://lists.ics.forth.gr/mailman/listinfo/crm-sig
--
--------------------------------------------------------------
Dr. Martin Doerr | Vox:+30(2810)391625
<tel:%2B30%282810%29391625> |
Research Director | Fax:+30(2810)391638
<tel:%2B30%282810%29391638> |
| Email:[email protected]
<mailto:[email protected]> |
|
Center for Cultural Informatics |
Information Systems Laboratory |
Institute of Computer Science |
Foundation for Research and Technology - Hellas (FORTH) |
|
N.Plastira 100, Vassilika Vouton, |
GR70013 Heraklion,Crete,Greece |
|
Web-site:http://www.ics.forth.gr/isl |
--------------------------------------------------------------
_______________________________________________
Crm-sig mailing list
[email protected] <mailto:[email protected]>
http://lists.ics.forth.gr/mailman/listinfo/crm-sig
--
--------------------------------------------------------------
Dr. Martin Doerr | Vox:+30(2810)391625 |
Research Director | Fax:+30(2810)391638 |
| Email: [email protected] |
|
Center for Cultural Informatics |
Information Systems Laboratory |
Institute of Computer Science |
Foundation for Research and Technology - Hellas (FORTH) |
|
N.Plastira 100, Vassilika Vouton, |
GR70013 Heraklion,Crete,Greece |
|
Web-site: http://www.ics.forth.gr/isl |
--------------------------------------------------------------
