I finally made the decision to release a version on TRNLTK. It is Turkish Natural Language Toolkit.

Here is some wiki about it: https://code.google.com/p/zemberek3/wiki/TRNLTK_Release_1_0_0

Here are the documents for TRNLTK : https://github.com/aliok/trnltk-java/blob/master/docs/102.md

I wrote a very robust morphological parser which is easy to maintain, extend and customize. I am proud of that. TRNLTK parser is an alternative to Zemberek, only proper open source project in Turkish language. A good working version of Zemberek morphological parser is implemented around 5 years ago and used in a lot of places, including major Linux distributions, major word processors and even big-browsers.

TRNLTK parser is solving the same problem IMHO in a much better way: morphologic parsing. Of course there are pros and cons. I wrote especially the cons in a little bit more detail in the wiki page I wrote above.

Again, even if this is a big problem for Turkish language because of its behavior, the real thing to solve is sentence parsing, part of speech tagging and the prerequisite of all, disambiguation.

Since January 2013, we work with the authors of Zemberek together and I hope we’ll find a good open source solutions to big problems described above.