About the model: Question 6
6. OWL contains the concept of “transitivity;” could it be an attempt to deal with inheritance?
“RDF (with its sub-standards included) has a long history, particularly from the DAML and OIL days (you may have heard of DAML+OIL), where the two different ontologies were merge quite successfully (and I suspect because each of them were quite beautifully designed) to form the basis for what RDF is supposed to be doing. Unfortunately, there were too many people and too much politics in the process, so the elegancy of DAML+OIL (which, really, was adequate for most modeling and were cleanly separated) turned into the 5 level RDF thing we’ve got now. They probably thought basic RDF was good enough to get you started, that more complex stuff should be handled by OWL (what does “more complex” mean in ontological terms?) and that the schema should express exchange rules (again, what does this mean in ontological terms?) The truth is that when doing ontology work the thing that makes them work is fuzziness and ambiguity, but the strictness and split personality of RDF is becoming a bit of a problem. I’ve seen some work on trying to bring these things together in an RDF 2.0 effort, but don’t know where they’re up to”–September 11, 2007 email from Alexander Johannesen.
I read a book on data modelling awhile ago that stated that object-oriented models support inheritance but relational models do not. Does RDF support inheritance?
“Yeah, it’s in the RDFS ;<rdf:RDF xml:lang=”en” xmlns:rdf=”http://www.w3.org/1999/02/22-rdf-syntax-ns#” xmlns:rdfs=”http://www.w3.org/2000/01/rdf-schema#”> <rdfs:Class rdf:ID=”Author”> <rdfs:comment>The class of people are authors</rdfs:comment> <rdfs:subClassOf rdf:resource=”http://purl.org/dc#Creator”/> </rdfs:Class></rdf:RDF>
It’s a sour mix of where in RDF you find what you can express, and how”–September 11, 2007 email from Alexander Johannesen.
(Do you agree with that characterization of object-oriented as opposed to relational models?)
“Sure, but I guess what they mean is that the relational model by default don’t carry inheritance constraints, and that’s very true, but it doesn’t mean the relational model can’t express them. In some ways you can define a lookup table as a form of inheritance mechanism, but it certainly isn’t a constraint nor documented as such, and, if one goes down to the SQL language, isn’t easy to express (but is certainly doable :)”–September 11, 2007 email from Alexander Johannesen.
I think inheritance is what we need to ensure that the cross reference from FBI to United States. Federal Bureau of Investigation will also apply when someone searches for the FBI Department of Counterterrorism. Do you agree?
“This is one of those biggies ; can we predict what context a user needs and / or want with any given search? Some times they indeed want the whole FBI as part of the search, at other times they’re just interested in the counter-terrorism part. Or, when we’re classifying, what parent class is that special zebra part of?”–September 11, 2007 email from Alexander Johannesen.
If so, can you help me figure out how to represent data in RDF such that inheritance is part of the model?
“Dig into the RDFS which deals with implications within the taxonomical part of RDF. I can’t provide much specifics per se as it’s been ages I’ve abandoned RDF as a modeling tool. I prefer to use Topic Maps (much cleaner as an expression engine) and then export it to RDF when need be”–September 11, 2007 email from Alexander Johannesen.DECISION: From my readings of RDF literature so far, I believe that by creating corporate subdivision as a subclass of corporate body and subject subdivision as a subclass of subject, the model supports hierarchy such that anything that is true of the main heading should be held to be true for the subdivision.