... unseen for a given domain s, we are alreadyperforming an implicit form of smoothing (whencomputing the expected counts), since each docu-ment has a distribution over all topics, and thereforewe ... Meeting of the Association for Computational Linguistics, pages 115–119,Jeju, Republic of Korea, 8-14 July 2012.c2012 Association for Computational LinguisticsTopic Models for Dynamic Translation ... appropriate.However, in both settings, 10 topics performed best.4 Discussion and ConclusionApplying SMT to new domains requires techniquesto inform our algorithms how best to adapt. This pa-per extended...