= Ontologies =

we need a "generic data ontology" with classes and properties that the adapters will usually return. These are classes like file, email, folder, appointment. and properies like title, fulltext, etc.

This main data ontology has to be extensible (at best, a protege project) and uploaded at a correct url.

we need one big RDF(S) or OWL lite ontology that contains all the nice triple predicate urls we want developers to use. This ontology has to be as straight as possible, clear namespace, downloadable file at the namespace uri, good documentation, according website, etc.

Chris: how about trying to use existing ontologies, e.g. such as found on [http://www.schemaweb.info/ SchemaWeb]? One problem with existing ontologies though is that often they don't exactly contain what you need (missing properties, too generic properties, no domain and range restrictions) that that you end up deriving your own ontology from it.

I also guess that the ontology will EVOLVE, but I hesitate to version it in the namespace (I would just add new classes/properties like foaf does). For changes in the ontology I would give the ontology a version number like 2005.1 2006.1 (the numbers you use in autofocus) but I would not change the namespace - this ususally brings only problems.

Chris: Where would this version number then end up? As a statement in the schema? Also I'm perfectly happy with regular 1.x.x style numbering, I do want this to look as application neutral as possible.

= special ontologies =
each datasource may bring a special ontology with it. How about a method getontology in ApertureDataSource?