pdf bibMinimum Error Rate Training by Sampling the Translation LatticeSamidh Chatterjee | Nicola CanceddaProceedings of the 2010 Conference on Empirical Methods in Natural Language Processing