Thanks for the update, Erik. I am in the process of designing my own Wikidata dataset and have some questions which I'll discuss with you offline.
There are several issues which I've found, though, that are of general concern so I'll post them here. All of this is based upon the WiktionaryZ tarball you released a couple months back, so sorry if some of this is no longer applicable:
1. Wikidata uses a single set of tables for storing all multilingual content: shorttext , translated_content . The problem with this approach is that it will not scale: small multilingual tables with limited ("seed data") content will be forced to do lookups on these gigantic and ever-growing tables, and exporting a single dataset will bog-down unnecessarily doing lookups (or even worse, full-table scans!) on these generic tables. I have put a page on Meta- [[m:Multilingual Wikidata]]- which discusses an approach for striping tables with multilingual content
2. The tables for defining languages and language groups seem to be shared between multilingual Mediawiki (the software) and WiktionaryZ (user data). I think this is not only conceptually incorrect, but also a security hole. See my comments from a while back: [[m:Talk:Ultimate_Wiktionary_data_design#.22Dog_Food.22]]
Anyway, it would be much appreciated if you could post more Wikidata designs as you come up with them. Thanks, and keep up the good work.