Re: Against Strong Ontological Commitment

On Friday, October 10, 2003, at 02:05  PM, Dan Connolly wrote:

> Why argue against a position that noone has taken?

Are you genuinely asking or indirectly suggesting I shouldn't?

Because it's not clear that  no one has taken it.

Also, it was involved a non-obvious consequence of the position, one, I 
believe, may affect some of the weaker views propounded or suggested. 
So I laid out the argument so we could understand better what might be 
wrong with SOC.

And if no one has taken it, what do you care that I chose to refute it? 
Perhaps I trying to build stronger consensus against it?

> What is "Strong Ontological Commitment"?

See the talking points for today email.

> If you want to argue against a position, do it in the
> words used to express the position, please.

I reserve the right to write my email in this forum in the context of 
this forum. This got sent not long after an email where I defined the 
term.

The OC view is:

"that use of a URI in RDF implies a commitment to its ontology, and if
there is doubt as to what ontology that is, the web may be used to
resolve it."

More precisely:

"that use of a URI in RDF implies a commitment to its ontology"

(The rest is irrelevant to the OC bit)

The strong OC view (SOC) is the interpretation of that phrase where:

	The 'implies' means that using *is* committing. If I so use, I so 
commit.
	The 'commitment' means that you must treat all the statements of that 
ontology as true.
	The ontology in question is the one the URI owner made dereferencable 
(or otherwise 		available)
I don't see much other likely readings, as I have argued. So long as 
the above text is "live" in a tag issue statement, I don't see arguing 
against it is either otiose or odious.

This statement of the view, in my understanding, entails that an RDF 
document is consistent iff the imports closure over all the URIs used 
in that document is consistent.

By RDF document being consistent, I mean that the a document of 
mimetype application/rdf+xml, conforms to the RDF specs, and that the 
RDF graph it is a serialization of is consistent.

"imports closure", in my lexicon, is the graph merge of the graphs 
serialized as documents which are dereferencable from the URIs labeling 
nodes in those graphs. Hmm. That didn't come out as clean as i wanted, 
but I *hope* we all understand that now.

(Note that I'm not saying that you "half to do the imports closure by 
downloading and parsing". But it would seem that it'd pretty dang hard 
to correctly say you've determined the consistency of a document 
without doing that.)

Have I met your terminological standards? And are you ok with me using 
SOC and OC (or their expansions) from now on without repeating this or 
referencing the archive?

Cheers,
Bijan Parsia.

Received on Friday, 10 October 2003 17:17:46 UTC