This paper explains how the GRUG Parallel Treebank has been produced.
These slides illustrate how the German-Georgian component of this dataset has been produced.