We are cutting away sample queries which do not add any insight. They
are slightly different but lead to similar conclusions. So it's better
to just have one.

I don't think I am going to use it as the table is way too big though.

Turns out: Honestly not that interesting. I think we can do better
with the evaluation.

As it appears many of the dcterms predicates remain unused. Even the
ones that do get used are used only very rarely.

Created using

 rdf4j-schema-generator/rdf4j-schema-generator -f application/xml \
   -c LOWER_UNDERSCORE -u http://purl.org/dc/terms/ -name dcterms \
   dublin_core_terms.rdf DCTerms.java

with the rdf4j-schema-generator [1] and the dcterms RDF file [2].

[1] https://github.com/ansell/rdf4j-schema-generator
[2] https://www.dublincore.org/specifications/dublin-core/dcmi-terms/dublin_core_terms.rdf

They are actually Q{1,2,3,7} from the original tetrapodal search paper.
It seems they are those queries that relate to organizational data.

Next up: Move to report, formalize.

We use the Extended Report which actually contains example
applications which is exactly what I will do next.

That's SPARQL and RDF4J.

- It's mostly about file structure/organization (duh)

- It also contains slightly deeper links, e.g. about on what objects
  other things are defined.

- In the end it's all about links. There are some data properties but
  they are honestly not that interesting. They are necessary for
  lookup however.