RE: [uos-convene] Technical feasibility of Interrelating upper ontologie

["Cassidy, Patrick J." <pcassidy@xxxxxxxxx>]

Jim,

   I can visualize several methods to find relations among the upper ontologies, including developing mappings (finding identical or near-identical classes and relations, or finding subsumption relations); trying to achieve a merger (which would require major changes in each of the upper ontologies); or trying to find a common subset.  Finding a common superset is another possibility, and may be required to realize the potential of category theory to formalize relations among ontologies.  Only the first technique, a partial mapping, is likely to be possible without changes in one or more of the existing upper ontologies, but that would leave some relations unresolved and some only probabilistic.  For a limited project, I think the common subset is both easiest and has the greatest benefits, though it also would require some modifications in one or more of the existing upper ontologies.

 

A discussion of some aspects of finding a common subset is on the Wiki page for the UOS:

      http://ontolog.cim3.net/cgi-bin/wiki.pl?UpperOntologySummit

 

If the custodians of the existing upper ontologies (at least the 3-D ones) agree to try to create such a common subset, and agree to make some modifications in their ontologies to create a subset that is logically compatible with each of the linked upper ontologies, I believe that a useful common subset ontology could be created fairly quickly, once funding becomes available.  The benefits I anticipate would be (1) to provide a solid reference ontology that can be referred to (as DC and FOAF are now) by people creating simple domain ontologies - permitting them, when they are ready, to convert to one of the more completely axiomatized ontologies without major changes or reinterpretations of their ontology elements; (2) provide a well-designed common reference ontology that will have some utility on its own to permit limited interoperability (i.e. interoperability using limited inferencing methods) and also permit those doing research on reasoning methods to compare results on the same substantial knowledge base - allowing them to change one variable at a time in their investigations, rather than the thousands when different knowledge bases are used for comparison; (3) provide a well-constructed ontology that can be learned more easily than the parent ontologies and used to develop some simple applications, providing valuable public examples of how a good ontology can be used; (4) perhaps the best part -- get a wide audience of budding ontologists accustomed to demanding meaningfully axiomatized relations for their ontologies, and increase the appreciation of the immense amount of careful detailed work that has already gone into the existing upper ontologies.  I would expect the demand for and use of the existing upper ontologies to increase as a consequence.

 

Any of the methods for interrelating ontologies are technically feasible, but those that require some modifications of one or more existing ontologies will only succeed if the custodians decide that the benefits are worth the effort.  That's not a technical issue, it's a cost-benefit analysis on the part of the custodians.  It is, of course, difficult to imagine a serious project emerging without some significant funding.  The IKRIS project (http://nrrc.mitre.org/NRRC/ikris.htm) appears to be tackling a similar though more limited problem, and their results (which may be available in April or May) could be a starting point for the development of the common subset.  Of course we can't predict exactly what the outcome will be.  If we could, there would be no need to engage all the intellectual firepower of the panelists who have agreed to take a serious look at the problem.  This is a first-ever try to solve a serious problem with this approach, and I hope that those in the community inclined to be skeptical for whatever reason can try to be supportive of the effort.  There are some questions that I expect will only be answerable after the panelists have had a chance to discuss the issues with each other directly.  This list and the Wiki are available for pre-meeting discussion, and I hope that the custodians will take the opportunity before the meeting to express their views.  But perhaps the higher-bandwidth face-to-face may be required to come to an agreement on what is possible and what isn't. 

 

Pat

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: uos-convene-bounces@xxxxxxxxxxxxxxxx [mailto:uos-convene-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Schoening, James R C-E LCMC CIO/G6
Sent: Monday, February 27, 2006 5:12 PM
To: 'Upper Ontology Summit convention'
Subject: [uos-convene] Technical feasibility of Interrelating upper ontologies

All,

        Are we going to discuss the technical feasibility of interrelating multiple upper ontologies?    To issue a communique, we'll need a good handle on whether we believe this is technically feasible and will enable semantic interoperability. 

James R. Schoening             
C-E LCMC CIO/G6 Office 
Voice: DSN 992-5812 or (732) 532-5812  
Fax: DSN 992-7551 or (732) 532-7551    
Email: James.Schoening@xxxxxxxxxxx