Michael, Kai, Daniel
Mostly discussions about the domain model again, only some notes were written down by Michael afterwards.
- 10:01 michaelp2 Hi Daniel, thanks for adding the definitions.
- 10:02 DGarijo no problem=20
- 10:02 michaelp2 I made a couple of changes to the domain model presentation based on your questions and Kai's responses.
- 10:03 michaelp2 Very minor stuff.
- 10:03 kai_ Ömer sent regrets
- 10:03 kai_ So we are complete
- 10:03 michaelp2 But I found it hard to explain in a concise fashion … I think I underestimated the time I needed there.
- 10:52 michaelp2 Topic: Cardinality refinement
- 10:55 michaelp2 Topic: Superclass of Description Set necessary? Domain/range problems in OWL, could be circumvented by property/chain inclusion?
- 11:04 michaelp2 Re Cardinality refinement: Annotation should be about exactly one description set, not 1..* to keep it inline with DCAM definitions (“descriptions are statements aboout ONE resource”)
- 11:09 michaelp2 … This precludes reuse of annotation to describe several descriptions sets, as this would require a new annotation. … for each description set instance. … In other words, reusing an annotation to describe a second description set makes it a new annotation automatically.
- 11:13 michaelp2 Re: Superclass of description set:
- 11:14 michaelp2 Kai: It might be necessary to assume a superclass of description set in our model, so that we can impose a range restriction on the “describes” relationship without OWL inferring every described entity being a description set. It might also be an aggregation of descrition sets.
- 11:16 michaelp2 Michael: But “describes” really is a bundle of relationships. Those could be lifted straight from the source vocabulary like “dct:creator”, so we don't get to impose domain/range restrictions at all. We got rid of a dedicated “provenance” relationship when we decided that we didn't want to connect annotation set with description set.
- 11:19 michaelp2 Kai: Could we do it another way with OWL, to state something like: Everything in an annotation set is an annotation, everything that is object of an annotation is a description set, ergo everything that is object of a relationship used in an annotation is a description set.
- 11:20 michaelp2 … used in an annotation set, I meant.
- 11:25 michaelp2 Michael: I think this would well be possible by using property chain inclusion in OWL 2: If x is value of a statement, and that statement belongs to an annotation set, x is a description set.
- 11:31 michaelp2 Michael: Or, in RDF terms: If resource x is object of property y, and property y is part of an annotation set, x is a description set. Hmm, on second look, doesn't sound right. I have to go and recheck. I think only one property can by propagated across another in a property chain inclusion.
minutes_2011_02_02.txt · Last modified: 2011/02/02 17:55 by kai