Applying Comparable Corpora to Machine Translation

The problem investigated here was how to improve statistical machine language translation between Polish and English speech. While excellent translation systems exist for many popular languages, it is fair to say that the development of such systems for Polish and English has been neglected. The most popular methodologies are not well suited for the Polish language and require adaptation. Polish language resources are lacking in parallel and monolingual data. Therefore, the main objective of the present study was to develop an automatic and robust Polish to English translation system to meet specific translation requirements and to develop bilingual textual resources by mining comparable corpora. Experiments were conducted mostly on casual human speech, consisting of lectures, movie subtitles, European Parliament proceedings, and European Medicines Agency. The aims were to rigorously analyze the various problems and to improve the quality of baseline systems, i.e., adaptation of techniques and training parameters to increase the Bilingual Evaluation Understudy (BLEU) score for maximum performance.