Best practice 14 – extracting features from text data