Output details

HMM word and phrase alignment for statistical machine translation

These algorithms are implemented in the open source MTTK Toolkit, which has been downloaded by students and researchers world-wide for teaching and university research. Cambridge translation systems using MTTK are competitive in international evaluations of machine translation: as noted for another selected output, the authors had the top-scoring Arabic-English system at the NIST (USA) 2009 Open Machine Translation Evaluation, and were amongst the top French/Spanish-English systems at the 2010 FP-7 sponsored Workshop on Machine Translation. The core algorithms described have attracted commercial interest: they have been reimplemented in a distributed, `big data', setting by SDL Research, Cambridge.