Benchmarking SMT Performance for Farsi Using the TEP++ Corpus

Statistical machine translation (SMT) suffers from various problems which are exacerbated where training data is in short supply. In this paper we address the data sparsity problem in the Farsi (Persian) language and introduce a new parallel corpus, TEP++. Compared to previous results the new dataset is more efficient for Farsi SMT engines and yields better output. In our experiments using TEP++ as bilingual training data and BLEU as a metric, we achieved improvements of +11.17 (60%) and +7.76 (63.92%) in the Farsi– English and English–Farsi directions, respectively. Furthermore we describe an engine (SF2FF) to translate between formal and informal Farsi which in terms of syntax and terminology can be seen as different languages. The SF2FF engine also works as an intelligent normalizer for Farsi texts. To demonstrate its use, SF2FF was used to clean the IWSLT–2013 dataset to produce normalized data, which gave improvements in translation quality over FBK’s Farsi engine when used as training data.