This document/repo is an exploration and various thoughts on how OWL might help describe data cubes and their dimension/dimension-values.
Due to the minor problems with the actual QB ontology I use a simplified version for illustration - and assume the existance of workarounds for the various uninteresting issues, and instead try to focus on the details.
There are two examples which both include OWL that works enough to demonstrate the possibilities for modelling.
The most comprehensive example (and in my mind the better one) is
sub-prop and is desribed in the document on materialising classes
on codes.
on-obs is likely I think an evolutionary dead end, but shows how we could materialise the classes onto observation objects themselves.
- Put the
bindirectory on your path and stardogsbindirectory on your PATH. - Run from the root of this directory
create-db sub-prop && create-db on-obs. This will build the stardog databases with OWL DL support. - You should now be able to run the queries and reason on the examples.
If you want to experiment with the data, you can run the command
refresh-db sub-prop which will drop and reload the data for you.
I used this crude workflow to develop and iterate the examples.
- Managing Codelists and Classifications with OWL
- 3 sector model
- Various problems doing this
- An existence proof for why modelling MECEness in OWL for statistical classifications might be hard/impossible; is that Ian Horrocks avoided doing it here