Investigating the Usefulness of Generalized Word Representations in SMT

We investigate the use of generalized representations (POS, morphological analysis and word clusters) in phrase-based models and the N-gram-based Operation Sequence Model (OSM). Our integration enables these models to learn richer lexical and reordering patterns, consider wider contextual information and generalize better in sparse data conditions. When interpolating generalized OSM models on the standard IWSLT and WMT tasks we observed improvements of up to +1.35 on the English-to-German task and +0.63 for the German-to-English task. Using automatically generated word classes in standard phrase-based models and the OSM models yields an average improvement of +0.80 across 8 language pairs on the IWSLT shared task.