RE: [BIONT] Teleconference

-Vipul,
 
> Discuss Alan Ruttenberg's use cases for BIORDF and identify relevant
artifacts (thesauri, ontologies, mappings,   
 >  etc.). 
 
Which use case you meant?  Or did I miss something?
 
 >  Brainstorm pragmatic and engineering definitions of ontologies in the
context of the use cases identified above .
 
I started contemplating this question and it seems not as clearly as I think
it would be.   If we give it a strict definition that an ontology is an RDF
model with an <owl:Ontology> header, the Dublin Core wouldn't be an Ontology
because "http://purl.org/dc/elements/1.1/" doesn't have an <owl:Ontology>
header.  The ontology header appears to serve mostly for annotation purpose.
Although the range of "owl:imports" is constrained to owl:Ontology, but it
is not clear from W3C's spec as what if the URI of an owl:imports is not an
owl:Ontology.  If an ontology can owl:imports an "ontology" like DC without
being labeled as some kind of error.  It implies that everything in RDF can
be considered as an ontology.  Although this definition is at first
seemingly absurd, there isn't a strong argument (at least I can not find
one) to rule against it. It appears to me, intuitively, that an ontology is
only an ontology when it is used by someonelse.  This relativeness seems in
line with its "conceptual" definition like - a spec of conceptualization.
In other words, some assertion only becomes an ontology when its
conceptualization is adopted by others.  But, how can we define it clearly
in the engineer sense? Anyone has any ideas?
 
Xiaoshu