[Top] [All Lists]

[ontac-forum] Reference model, domain knowledge and ontological analysis

To: <ontac-forum@xxxxxxxxxxxxxx>
From: "Gary Berg-Cross" <gary.berg-cross@xxxxxxxx>
Date: Tue, 1 Nov 2005 09:40:40 -0500
Message-id: <330E3C69AFABAE45BD91B28F80BE32C905626D@xxxxxxxxxxxxxx>

I had one perspective I wanted to contribute to the discussion about “parts” between Nicholas and Barry.  The exchange was:


N>We need identity to make sense of "a123 part-of b456". How do I

N>recognize 'a123' among all possible instances?  


B>Typically it will be you who has baptised the relevant instance by

using this designation. Thus you may have baptised your heart as

'a123'. If you do not know what instance 'a123' stands for, then do

B>not use this instance designator in your work!   


Perhaps part of Nick’s question concerns the cognition of “recognition” and how it relates to our ontological assertion.  Barry describes the “designation” as having been “baptized”, which I take to be a certain type of formalization.  It seems to me that one interpretation of the cognitive processes involved in ontological analysis (OA) is that it is model based.  The analyst employs a “reference model”, which in might a common sense one of parts, but for sophisticated analysis should be an externally available idea such as from mereology.  If I am a beginner in OA, I might not use many distinctions and indeed may not understand how proper parts (proper part, PP, is any part excluding the whole ) or components differ from fiat parts (e.g. the engine is a component of a part, while the left side of the car is a part by fiat).


Which brings me to another part of the exchange on an approach to avoidling problems


N>I have a feeling that, at a coarse level, we're more or less in agreement.

>At a practical level, there are wholes that are problematic w.r.t what

N>should we be doing next.    (09)


B>Don't let worries about the problematic cases keep you from doing

B>good work with the non-problematic cases.   



What’s a problem for one modeler may not be one for another one who knows more about a domain and/or ontological principles.  Different modeling judgments (domain interpretations)  give us “conceptualization mismatchs.  To overcome this we want a common reference model on semantic relations.  For example,  something that Barry would be very familiar with, M.E. Winston, R. Chaffin, and D. Herrmann.in “A Taxonomy of Part-Whole Relations.” (Cognitive Science, 11:417–444, 1987) identifyed six  meronymic senses of part-of in of underlying the semantics of English usage. The senses are: component-integral, stuff-object, portion-mass, place-area, member-collection, feature-activity.


The GALEN ontology of human anatomy, for example, uses the first 5 of these as a reference and there is a formal analysis or each.  But this formalism, available in the reference model, must be “recognized” and correctly applied (a cognitive activity) by the ontological analyst.  So U. Hahn, S. Schulz, and M. Romacker.in “Part-whole reasoning:

A case study in medical ontology engineering. IEEE Intelligent Systems, 14(5):59–67, 1999 translated the anatomical terms in UMLS into description logic.


 In such work a single anatomical entity is modeled by additional concepts that denote the structure of the entity and their current understanding of the  set of parts that correspond to the entity – this may change over time and be re-baptized for various reasons. A property such as perforation-of attributed to Colon Structure may correctly generalize to Intestine Structure, as these entity structures are in an is-a/component relationship.  A property which Hahn & Schulz suggest an analyst might make a mistake  on is “inflammation-of” which they interpret as  defined to apply to an organ entity, but not the entity structure of the organ entity,  Well this is based on the analyst’s domain knowledge which gets formalized in the Ontology.


So an inference I would make (going back to an earlier question of mine) is that while a completed ontology might avoid the issues of language and knowledge, the process of developing an ontology runs into both of these – which makes developing a common ontology hard.  Not to say that it is not good to start with the low hanging fruit, just to add that the group needs to have common reference models and common techniques of its own and access to good domain knowledge.



Gary Berg-Cross





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>