It would be nice to have a super TM (translation memory) which unifies all the TMs around in the world. There are thousands of them and most of the translaton work is basically done with very high quality if all the TM systems could unify.
Topic on Talk:Content translation
That would be lovely, although not planned at the moment. I hope to plug much more machine translation and translation memory services some day, but it will happen step by step. Thanks for the suggestion!
I think if Wikipedia leads the way most will follow. It is a big responsibility for wikipedia. I would be willing to help as much as possible in this project. It can be done and should be done together with machine translation. Both are very compatible, the more good translations you give to the machine translation the better are the results. The task of translators and revisors would be just to make small corrections. You could also use this platform, which would be hugh to get money, because many translators all over the world would use it to make their work and most would be willing to pay for it.
Furthermore you could pay some really good professional dictionaries, like the ones in acolada and similar platforms to give them to the translators.
You can collaborate more close with www.proz.com and other similar platforms with terminology work.
Thanks a lot for these ideas!
Do you mind mentioning it also at meta:Talk:Strategy/Wikimedia movement/2017/Cycle 2/A Truly Global Movement? (See also meta:Strategy/Wikimedia movement/2017/Cycle 2/A Truly Global Movement?
(edit conflict) Hello, as I understand you want a tool that centralize translations systems like Google Translate.
ContentTranslation, the tool edited for Wikipedia and others, was make to help translate wikis pages only.
I think that your proposition is not particularly concern Wikimedia Foundation, but can be a good project to help translate other documents and not only wikis pages.
Have a nice day
Google's neural net translation on French-English is getting better now, and should be considered for inclusion into the Content Translation system. Perhaps letting people purchase their own Google Translation API key and then having them upload their API key to Wikipedia in their preferences. From then on, Content Translation could use Google Translation for the machine translation. In this way the commercial expense would be paid by the translator and not by Wikipedia. The neural net translation system is a step better than the statistical based approach they used earlier, and is now state of the art machine translation.