Unknown Words Mining

We describe a data selection method for domain adaptation in machine translation, based on relative frequency ratios computed between in-domain and out-of-domain corpora. Our method is compared to a state-of-the-art …...