Morphological Constraints for Phrase Pivot Statistical Machine Translation
This work addresses the challenge of limited parallel data for low-resource language pairs in SMT, but it is incremental as it builds on existing pivoting techniques with morphology constraints.
The paper tackled the problem of low-quality translations in phrase pivot statistical machine translation (SMT) when using a poor morphology language as a pivot between rich morphology languages, by using synchronous morphology constraint features, resulting in a gain of 1.5 BLEU points over a phrase pivot baseline and 0.8 BLEU points over a system combination baseline for Hebrew-Arabic SMT pivoting on English.
The lack of parallel data for many language pairs is an important challenge to statistical machine translation (SMT). One common solution is to pivot through a third language for which there exist parallel corpora with the source and target languages. Although pivoting is a robust technique, it introduces some low quality translations especially when a poor morphology language is used as the pivot between rich morphology languages. In this paper, we examine the use of synchronous morphology constraint features to improve the quality of phrase pivot SMT. We compare hand-crafted constraints to those learned from limited parallel data between source and target languages. The learned morphology constraints are based on projected align- ments between the source and target phrases in the pivot phrase table. We show positive results on Hebrew-Arabic SMT (pivoting on English). We get 1.5 BLEU points over a phrase pivot baseline and 0.8 BLEU points over a system combination baseline with a direct model built from parallel data.