Aldo Gangemi wrote: > Dear CRM specialists, I'd like to point you at a new OWL version of > CIDOC 4.2, which I have produced for other purposes. It is based on the > official RDFS version, and besides the semantic translation, it only > includes a guess about the datatypes used in some CIDOC properties. > Please refer to file documentation for details (I've put it in a ftp > area of my lab, but if you find it useful, please copy it where > appropriate: > > http://www.loa-cnr.it/ontologies/CIDOC/cidoc_v4.2.owl
I like the idea of expressing alignments between ontologies using OWL constructs such as equivalentClass. However, a straight automatic translation from RDFS to OWL misses the opportunity to formalize some of the (textual) definitions from the CIDOC CRM that cannot be expressed in RDF Schema. Wouldn't it be useful to include some more definitions (explicit and implicit) from the reference model? As an example, pairs of inverse properties are identified by the letters B and F appended to the numerical part of the property name. This is mnemonics for humans, but not easily processed by machines. However, it can easily be made processable by using the owl:inverseOf statement. There are further examples such as the disjointness of some classes (explicitly mentioned in the text) or the transitivity of some properties (e.g. P120F.occurs_before). One could go even further and declare P57F.has_number_of_parts as owl:FunctionalProperty, assuming that no instance of E19.Physical_Object can consist of different numbers of parts at a given time. How does the SIG think about a coordinated effort that would eventually result in an official OWL representation of the model? Some work has already been done and I think it shouldn't be too hard to reach a consensus on what to express in OWL language constructs, and what to leave out. Best wishes, Detlev
