FWIW, I have implemented a query-able stand-alone web
server that keeps all
of the wikidata property-item-links in memory. This uses the wikidata dumps
That does not sound too terribly scalable. I did the same thing
(custom webserver, data kept in memory) for the map labels of my
WikiMiniAtlas, and had to abandon this in order to be able to support
more languages. And my data is only a subset of what I expect to show
up in WikiData.
On top of that not being able to join data from there with other DBs
is a serious deficiency. Same goes for the suggestion to "just use the
API".
Daniel