You can read all about it in our paper which has just appeared in Association for Computational Linguistics and can be read free of charge here.

In line with the Open Data policy of our project, the data and code are on our GitHub page.