RE: LSIDs and ontology segmentation

Dear Chimezie,

> I would think that an author of an ontology of this size 
> would want to consider fragmenting the ontology (perhaps by
> sub-domains) and linking them with owl:imports.  In such a 
> scenario, the terms could simply be identifiers asserted 
> within each ontology fragment and only the ontology fragments 
> would need URLs for dynamic resolution.

We are on the same page.  But I would disagree with youu on the usage of
owl:import.

The help of owl:import is only partial.  For instance, 

If o2 owl:imports o1, o2 is tightly bound to o1.  If I want to develop
another ontology say, o3, and want to use the concepts of o2, I must also
have to import the o1 as well.  

First, such kind of chain dependency will eventually leads to a
big-monolithic ontology.

Second, it hurts the sharing capability of o2.  What if I only want to
import o2 but not necessarily agree to o1?  I must either pay the price,
i.e, to be forced into using o1. Or impossible because I might have some
assertions in o3 that might incur inconsistency with o1 but not necessarily
o2.

In my proposed methodology of ontology normalization (some minimum
description at my website
http://www.charlestoncore.org/ont/2005/08/o3.html).  You would create o1 and
o2, each of which only make assertions of the term developed under their
respective namespaces.  Then, you create a profile, say p1, where you map
your usage of the imported ontologys.

By this way, if a user agrees with you on your alignment from o2 to o1, he
just simply import p1.  But for those who don't, they can still create a
profile to use the o2 without being forced to take o1.

This is only one form of normalization (intension normalization in my
terminology). It can help make ontologies to be piecemeal built and
gracefully evolve.

There is an more extreme form of normalization (extension normalization)
that will help to maintain the lexical stability of the URI and can handle
the reolutionary change of an ontology.

Xiaoshu

Received on Thursday, 13 July 2006 19:44:13 UTC