CatBert on F{}OWL Ontologies (S01E02)

CatBert was sitting in front of his CatBook. He again made this hissing sound letting me know that he is onto something.

"Ok, what is it today?"
"Uhm, no. What about it?"
  • "Well, this guy is observing that many Semantic Web organisations, conferences, companies, etc. have not put their own content into RDF form. And then he quotes the statements, or shall I say excuses why this has not been done yet."
"Ok, but RDF is not even 10 years old. How should we expect that SW evangelists eat their own dog food so quickly?"

CatBert gave me the look, but did not pick up on my humorous attempt to ease his criticism.

"That's good, not?"

Oh oh, I thought. Now I knew where this was heading. And indeed CatBert had already loaded the ontology into Protege.

  • "There are a few problems with that."

I tried to sound as desinterested as possible. But there is no way stopping CatBert bashing innocent ontologies.

  • "The ontology cannot be loaded over the Internet as the author has used relative links somewhere."
"Ok, but that is easy to fix, right?"
"Ok, but that is easy to fix too."
  • "Sure, but then there are number of strange declarations:"

:Program_Chair rdfs:subClassOf :Organizing_Committee .
:General_Chair rdfs:subClassOf :Organizing_Committee .
:Panel_Chair   rdfs:subClassOf :Organizing_Committee .

"Ok, that probably doesn't make much sense. How can a chairman be a specialized committee?"
  • "It gets better: There is also this:"

:ProgramChair a :Program_Chair .
:GeneralChair a :General_Chair .
:PanelChair   a :Panel_Chair .


Admittedly, I found this rather odd too. At some point one has to decide whether something is modelled as individual or class. But not both.

  • "It gets better: There is also this:"

:Responsibility rdfs:subClassOf :RuleML-2008 .

"Right, but I know that one. That just means that :RuleML-2008 is meant as a placeholder for Concepts. That's not serious."

CatBert hissed again. His way to let me know he did not buy the argument.

  • "And what about this?"

:ProgramChair a :Responsibility .
:GeneralChair a :Responsibility .

I pressed my lips, trying to think of a universe in which that would make any sense. But before I could come up with something, CatBert continued:

  • "The fun starts when you look at the object properties: accountable, consulted, informed, responsible, signs, etc."
  • "They are all made functional. So if someone is on one committee he or she is responsible for exactly one responsibility."
"Sounds logical to me."
  • "Not really, human."

CatBert really knows how to drive home an insult.

  • "Functional means that along that property there can only be one value. If there are more, a reasoner would glue them together. What one wants here is inverse functional so that for one responsibility there is exactly one person responsible for it."
  • "You should know that. You are giving a Semantic Web course at an university here."
"Ok, ok, that ontology is probably not really completely thought through."
  • "Looks that way. And if one really draws up that many similar properties, then thinking whether one is a specialization of the other is certainly a service to catkind. Or humankind in your case."

I sighed. CatBert had a point. Time to feed him with his favorite food.

angry-cat-small.jpg857 bytes
RuleML-2008.owl6.81 KB
Posted In


Loved your protagonist. I wish there were some good pictures of him / her / it.

And your point is poignant; if people who should be in the know model crap, what we get is crap, and what it means is that modeling is hard, which means that no matter how cool the technology, it will only end in tears.

Better reasoning with RDFS, then? *grin*

Alex (not verified) | Fri, 02/06/2009 - 20:35

Re: Hahaha

Better reasoning with RDFS, then?

The problem I notice is that a few people do not even get the classes thing right. So RDFS, RDFS+ or RDFS++ will suffer all the same.

Nota bene: I'm not saying that thinking in classes is actually the right thing to do. But when you go with OWL, and the only thing it really gives you is subsumption then it is pretty stupid to throw that out of the window.

rho | Sat, 02/07/2009 - 20:46


Amen, brother! If the framework demands the notion of classes, then if the people don't get classes you're stuffed. Lucky Topic Maps deal with types, eh? :)

This is actually a much more serious problem than a lot of people think it is, a question of philosophy and formal definitions within information science, no less. And I am not sure classes are the right answer, either. On a similar note, I've thrown tantrums about object-oriented programming before, that it unfortunately looks so much like truth, people start treating it as truth. Very annoying.

I wish someone could come up with something more fuzzy. All these formal definitions lodged in logic is killing the potential.

Alex (not verified) | Sat, 02/07/2009 - 22:44

CatBert to the rescue

CatBert to the rescue :)

Thanks for that.

squig (not verified) | Sat, 02/07/2009 - 14:33