Actually it is impossible to collect all achieved results in one single article. Compared to all other events we went to, this was the most productive one and we will need months to actually work on all that will need to be done. Let's try to point out what interests our users most: language learning.

Up to now we only considered Parley and KHangman, well ... time to broaden our list of “interesting software” with learning Alphabets: Klettres is the new bit we decided to add. But for now let's center on Parley.

During the past months we collected data with means of tables and finally, just a couple of weeks ago Ambaradan went online. Right now we still don't edit there, but it is only a matter of time. What we already can do is provide data. The sprint gave us the possibility to run a test of for applications: Parley and Ambaradan based on terminology by FAO: Agrovoc, which thanks to its extensiveness is a great test-object. There are between 80.000 and 100.000 entries in 8 languages. Parley needs kvtml2 format, that is a particular xml format. During the sprint an export script for Ambaradan to directly produce kftml2 format was created and worked just fine. Parley had to digest a really huge file with 100.000 entries (source/target language pairs) and it did really well. The speed remained good, no slow-down.

One caveat are the lessons: these still have to be created for Agrovoc, therefore we have quite some work waiting :-)

But what does all this above here mean? Well: Parley will be able to retrieve data directly from Ambaradan, without intermediate steps. This means that we can concentrate our efforts on the data which of course means more and better quality. So happy language learning with Parley and the other applications working with kvtml2.