Statistics for Improving the precision of example-based machine translation by learning from user feedback