Edinburgh’s Syntax-Based Systems at WMT 2014

Proceedings of the Ninth Workshop on Statistical Machine Translation, pages 207–214, Baltimore, Maryland USA, June 26–27, 2014. c⃝2014 Association for Computational Linguistics Edinburgh’s Syntax-Based Systems at WMT 2014 Philip Williams1, Rico Sennrich1, Maria Nadejde1, Matthias Huck1, Eva Hasler1, Philipp Koehn1,2 1School of Informatics, University of Edinburgh 2Center for Speech and Language Processing, The Johns Hopkins University Abstract This paper describes the string-to-tree sys- tems built at the University of Edin- burgh for the WMT 2014 shared trans- lation task. We developed systems for English-German, Czech-English, French- English, German-English, Hindi-English, and Russian-English. This year we improved our English-German system through target-side compound splitting, morphosyntactic constraints, and refine- ments to parse tree annotation; we ad- dressed the out-of-vocabulary problem us- ing transliteration for Hindi and Rus- sian and using morphological reduction for Russian; we improved our German- English system through tree binarization; and we reduced system development time by filtering the tuning sets. 1 Introduction For this year’s WMT shared translation task we built syntax-based systems for six language pairs: • English-German • German-English • Czech-English • Hindi-English • French-English • Russian-English As last year (Nadejde et al., 2013), our systems are based on the string-to-tree pipeline implemented in the Moses toolkit (Koehn et al., 2007). We paid particular attention to the production of grammatical German, trying various parsers and incorporating target-side compound splitting and morphosyntactic constraints; for Hindi and Rus- sian, we employed the new Moses transliteration model to handle out-of-vocabulary words; and for German to English, we experimented with tree bi- narization, obtaining good results from right bina- rization. We also present our first syntax-based results for French-English, the scale of which defeated us last year. This year we were able to train a sys- tem using all available training data, a task that was made considerably easier through principled filtering of the tuning set. Although our system was not ready in time for human evaluation, we present BLEU scores in this paper. In addition to the five single-system submis- sions described here, we also contributed our English-German and German-English systems for use in the collaborative EU-BRIDGE system com- bination effort (Freitag et al., 2014). This paper is organised as follows. In Sec- tion 2 we describe the core setup that is com- mon to all systems. In subsequent sections we de- scribe language-pair specific variations and exten- sions. For each language pair, we present results for both the development test set (newstest2013 in most cases) and for the filtered test set (new- stest2014) that was provided after the system sub- mission deadline. We refer to these as ‘devtest’ and ‘test’, respectively. 2 System Overview 2.1 Pre-processing The training data was normalized using the WMT normalize-punctuation.perl script then tokenized and truecased. Where the target lan- guage was English, we used the Moses tokenizer’s -penn option, which uses a tokenization scheme that more closely matches that of the parser. For the English-German system we used the default Moses tokenization scheme, which is similar to that of the German parsers. For the …

Content truncated to 500 words