Билексическая статистика

We present a semi-supervised approach to improve dependency parsing accuracy by using bilexical statistics derived from auto-parsed data. The method is based on estimating the attachment potential of head-modifier words, …...