RE: LC Comment: "Hidden" Axioms

The mention of "superseding" and "superseded" begins to sound like
incremental versions.  Might one want more than just a list of what is
"hidden" and what is "active"?  I would think that much of the value of
the non-active axioms would include, at bare minimum, the relative order
in which new axioms were introduced and "hidden" axioms were
deactivated.

It would be possible to introduce new axioms via the notion of additive
"micro-versions" - incomplete collections of axioms that, when combined
with a previously complete ontology, creates a new version of the same.
This still presents the same fundamental problem - how to indicate that
an axiom should no longer have logical force - but it would allow the
relative order of changes to be encapsulated in the information
associated with the "micro-version" itself, something that is available
already.

Let me mention related requirement that we have encountered.  We have a
need to be able to classify an ontology and subsequently transmit both
the asserted and the logical inferences for display and consumption in
secondary resources such as wikis and other tools.  As these tools may
create additional axioms, we need to differentiate the asserted from the
inferred - both as important information to the editors and to be able
to remove or ignore these axioms when the modified ontology is
subsequently re-classified. While this is a slightly different use case,
it still involves the same notion - some sort of tag or property on an
axiom that affects the way that it is interpreted. 


Harold Solbrig | Division of Biomedical Statistics and Informatics |
507-293-3774 | solbrig.harold@mayo.edu

-----Original Message-----
From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
On Behalf Of Bijan Parsia
Sent: Friday, January 23, 2009 3:33 PM
To: W3C OWL Working Group
Subject: LC Comment: "Hidden" Axioms


This is the result of discussions with people at Manchester working on  
a collaboration with Siemens Medical Health Services (i.e., an  
industrial project).

We would like the ability to "turn off" the logical force of axioms  
without having to remove them from the structural model. Currently,  
the only way to directly hide an axiom from a reasoner in an OWL  
document is to use one of the commenting capabilities of the  
serializations (e.g., <!-- -->). However, this isn't really workable  
for a number of reasons including that such comments tend to not  
survive conversion, or even parsing (since it has no place in the  
structural model) and such comments are difficult to retrieve or  
manipulate.

One basic use of this is to maintain superseded (or potentially  
superseded) axioms in the same ontology as the superseding ontology.  
In a sense, this is like deprecatedClass except that it has a clear  
operational meaning (whereas deprecatedClass was unspecified). From an  
API point of view, hiding an axiom (rather than deleting it) is a  
natural way of exploring the consequences of changes. It's nice to be  
able to persist such experiments in the canonical format.

Ideally, in our view, this would be effected by a designated  
"annotation". I've gotten some feedback that this would be an  
exception to the "annotations don't affect the meaning".

True. But I don't think it's a harmful exception *if built in*. For  
obvious reasons it *can't* be conformingly implemented as an extension  
to OWL 2 since all user defined annotations can be deleted without  
changing the logical meaning of the ontology. Keeping it as an  
annotation (rather than a distinct syntactic flag) is both easier to  
spec (since it doesn't change as many documents) and can be combined  
with axiom names to produce multiple external "profiles" of an  
ontology (e.g., by hiding different sets of axioms from different  
external files which import the core ontology). This makes it easier  
to make cut down versions of an ontology from the same source.

I apologize for not raising it earlier. I had been thinking that it  
could all be handled at a higher level. But in order to publish  
ontologies with hidden (in this sense) axioms, it has to be part of  
the core spec. Otherwise, conforming tools will misinterpret such  
ontologies.

Cheers,
Bijan.

Received on Friday, 23 January 2009 22:46:39 UTC