[Top] [All Lists]

[ontac-forum] RE: [dolce] A question about collections

To: <dolce@xxxxxxxxxxxxxxx>, "Cassidy, Patrick J." <pcassidy@xxxxxxxxx>
Cc: g.steve@xxxxxxxxxxx, d.pisanelli@xxxxxxxxxxx, ontac-forum@xxxxxxxxxxxxxx
From: "Obrst, Leo J." <lobrst@xxxxxxxxx>
Date: Thu, 8 Dec 2005 17:49:52 -0500
Message-id: <9F771CF826DE9A42B548A08D90EDEA80A53A7B@xxxxxxxxxxxxxxxxx>
I'd be interested in the paper too.
Dr. Leo Obrst       The MITRE Corporation, Information Semantics
lobrst@xxxxxxxxx    Center for Innovative Computing & Informatics
Voice: 703-983-6770 7515 Colshire Drive, M/S H305
Fax: 703-983-1379   McLean, VA 22102-7508, USA

From: dolce@xxxxxxxxxxxxxxx [mailto:dolce@xxxxxxxxxxxxxxx] On Behalf Of Aldo Gangemi
Sent: Thursday, December 08, 2005 4:02 PM
To: Cassidy, Patrick J.; dolce@xxxxxxxxxxxxxxx
Cc: d.pisanelli@xxxxxxxxxxx; g.steve@xxxxxxxxxxx; ontac-forum@xxxxxxxxxxxxxx
Subject: RE: [dolce] A question about collections

Dear Pat, in 1997-1999 the current Rome group of LOA (Ontology and Conceptual Modelling Group of ITBM-CNR at that time) had made a UMLS-SN alignment (actually not only an alignment, because several strategies have been applied to import "types", "relationships", informal "templates", and to axiomatize the glosses) to an ontology (ON9) partly similar to DOLCE.

The original Ontolingua code including ON9 and the medical plugin based on the UMLS-SN alignment are at this URL: http://www.loa-cnr.it/medicine/.
Some publications about that work as well as the re-engineering of the Metathesaurus can be downloaded from the LOA site.

Consider that a previous version of ON9 had been used by Teknowledge in the development of SUMO.

I've considered porting ON9 medical section under DOLCE, so if someone is doing the work, that can be highly facilitated by looking at what we have done. We could also provide some assistance.


PS a copy of the paper separately to you

At 19:09 -0500 22-11-2005, Cassidy, Patrick J. wrote:
  I would also like to get a copy of your paper.
  The ONTACWG I mentioned previously (http://colab.cim3.net/cgi-bin/wiki.pl?OntologyTaxonomyCoordinatingWG) is discussing a group effort to compare upper ontologies by doing a formalization of the UMLS Semantic Network with respect to each of the upper ontologies of interest to members -- OpenCyc, DOLCE, SUMO, BFO, ISO 15926.  The suggestion is that by trying this kind of formalization on a restricted size ontology (The UMLS-SN had 1230 classes and 54 semantic relations), a volunteer group like the ONTACWG might be able to collect useful information on just how different these upper ontologies are and whether a merger is feasible.  We still have no funding.  I do hope that if we can make some meaningful progress on this limited effort, we may be able to attract funding from some source.
  If anyone on the DOLCE list knows of similar efforts, I would appreciate being kept aware so as to avoid duplicative effort.  Our effort is, as I mentioned, is fully open to participation from any source, and the results will be public domain or freely usable with attributions of sources (e.g. Cyc or SUMO or DOLCE).

Patrick Cassidy
MITRE Corporation
260 Industrial Way
Eatontown, NJ 07724
Mail Stop: MNJE
Phone: 732-578-6340
Cell: 908-565-4053
Fax: 732-578-6012
Email: pcassidy@xxxxxxxxx


From: dolce@xxxxxxxxxxxxxxx [mailto:dolce@xxxxxxxxxxxxxxx] On Behalf Of Aldo Gangemi
Sent: Tuesday, November 22, 2005 6:46 PM
To: dolce@xxxxxxxxxxxxxxx; Gilles Kassel
Cc: oltramari@xxxxxxxxxx; ferrario@xxxxxxxxxx; catenacci@xxxxxxxxxx
Subject: Re: [dolce] A question about collections

Hi Gilles,

thanks a lot for your remarks. We were aware of the problem, because we had assumed constitution in a different, larger meaning at the time of the report you are referring to.
We have now a different solution, implemented in a new article, which does not use either dependency or constitution from DOLCE. Instead, it uses cognitive schemas as an intuitive basis for introducing membership.
I'm sending you the article separately. For those in the DOLCE list that are interested too, please contact me.

Some more comments inside.

At 9:23 +0100 21-11-2005, Gilles Kassel wrote:
Dear all,
I read with great interest the paper entitled "From Collective Intentionnality to Intentional Collectives: an Ontological Perspective" (a submitted paper available on DOLCE's publications page). Actually I would like to reuse this conceptualization to reconstruct an ontology of organizations that we recently proposed at LaRIA (Amiens, France) with a PhD student. My objective is also to integrate this conceptualization in the ontological resources of the OntoSpec methodology (see the technical report about OntoSpec on DOLCE's site). However, the way the first layer of the proposal is defined, that is the notion of collection, seems to me raising a problem, notably in its articulation with DOLCE.

On one hand, it is recalled that in DOLCE « relations between instances of the same category are admitted, such as part, constitution, connectedness, etc." And, indeed, the axioms Ad21 and Ad22 published in (Masolo et al., 2003) say that "physical (respectively non-physical) endurants constitute, and have for constituents, only physical (respectively non-physical) endurants".


In this paper, on another hand, collections are defined as social objects, therefore as non-physical endurants, and the membership relation as a constitution relation. And, in the same time, it is claimed that physical endurants can constitute a collection: "Endurants constituting a collection are either mereotopologiccaly unconnected (e.g., statues in a statuary) or weakly connected (e.g., a pile of plates)".
Therefore, if the constitution relation is the same as DOLCE's one, and I guess it is the case, there is clearly a contradiction.

Correct, as I was saying, we had assumed constitution in a more tolerant sense, dealing with so-called "stratification", so that social entities can "emerge" out of physical ones. The current restriction for "generic constitution" in the OWL version of DOLCE-Lite-Plus is indeed different from the S5 version of full DOLCE.
In an attempt to fully axiomatized the tolerant notion of constitution, we have encountered some interesting properties of emergence: for example, it seems that constitution can be kept compliant with DOLCE axioms if we assume that the emergent collection is constituted by the *information content* of the members, and not directly by the members. An appropriate chain of relations allows to refer to members anyway.
Anyway, such notion of constitution needs more work, and after all, does not make justice to the "basic" intuition of a collection, one the most important patterns acquired during cognitive development. For the time being we then decided to adopt a cognitive schema, formalized in an appropriate way (see article).

In a first attempt to extend DOLCE to build an ontology of organizations, we tried to reuse the notions of agentive-group and unitary-collection of (Masolo et al., 2003) defined as physical objects. However, we met some problems: i) firstly, the lack of a generic notion of collection and the difficulty to propose one ; ii) secondly, the fact that when members of a group (or a collection) are geographically dispersed, considering the whole as a physical object seems not tenable ; iii) lastly, the axiom Ad29, relating the part and constitution relations, seems too strong if we interpret a sub-group (or a sub-collection) as being a part of a group (or a collection): indeed, if x is a member of a group (or a collection) y and y' is a sub-group (or sub-collection) of y not containing x then no x' being part of x and constituting y' will existŠ

The proposition made in this paper has (from my opinion) some advantages, in particular concerning the points i) and ii) above. However, I guess that it contradicts the axiom Ad22. Moreover, I still have problems concerning the axiom Ad29. If, for instance, the technical staff and the commercial staff of a society are considered as parts of the staff, then, for the reasons mentioned above, it is easy to find counter-models for Ad29.

Ad29 is also at odds with the tolerant notion of constitution (not exactly for the reason you indicate btw), in fact, it would imply that, if Gilles is a member of a staff, its leg should be a constituent of a part of that staff. And this makes no sense. As I said, we were thinking about a Hartmann-like notion of emergence, not about DOLCE's one, but this fact is wrongly described in the old report. In the new one, consitution is no more used.

I take advantage of this mail to raise a last a last question. The fact of defining collections as social objects, and not as non-physical endurants in general, excludes private collections that would be embodied in physical agentives as mental objects. Is there a particular reason for this exclusion?

Mental objects are underdeveloped even in the DOLCE-Lite-Plus version. We are actively working on integrating social and mental objects. BTW, I don't see any problem in having collections of mental objects, despite the fact that collections are social objects.
By "social" we mean something that can be communicated (in principle), and relies on some encoding. By "mental" we try to catch something that is not necessarily communicable. But a collection is by definition communicable, because it is a reification of an (extensional) class, which is an abstract entity, then it stands on an encoding system.
One could wonder if there exists something which is collective but exclusively mental, i.e. non-communicable. We haven't gone so far yet. We are still investigating on the possibility that there are mental objects as a disjoint class from social ones. If one decides not, your problem disappears.
Notice that we are not trying to be prescriptive on what is mental or not, but we need to understand on which grounds something can be claimed to be mental without being social (and of course without being a physical neurobiological process or object). Philosophy of mind has spent substantial resources in the attempt to solve the problem ...
I suppose Alessandro Oltramari, Roberta Ferrario, and Carola Catenacci have probably more beef about these issues.

Still thanks for the great questions.


Aldo Gangemi
Research Scientist
Laboratory for Applied Ontology
Institute for Cognitive Sciences and Technology
National Research Council (ISTC-CNR)
Via Nomentana 56, 00161, Roma, Italy
Tel: +390644161535
Fax: +390644161513

Science fair projects Science projects Computer science Dolce



Aldo Gangemi
Research Scientist
Laboratory for Applied Ontology
Institute for Cognitive Sciences and Technology
National Research Council (ISTC-CNR)
Via Nomentana 56, 00161, Roma, Italy
Tel: +390644161535
Fax: +390644161513

Science fair projects Science projects Computer science


Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki: 
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG    (01)
<Prev in Thread] Current Thread [Next in Thread>
  • [ontac-forum] RE: [dolce] A question about collections, Obrst, Leo J. <=