Context Based Machine Translation With Recurrent Neural Network For English - Amharic Translation

The quote from Rev. Jesse Jackson, A text without a context is a pretext",rnsummarizes the reasoning behind this thesis. Capturing context in translatingrnbetween two human languages using computing machines is challenging.rnIt is more challenging when the languages di er greatly in grammar and havernsmall parallel corpus like the English-Amharic pair. The current approachesrnfor English-Amharic machine translation usually require large set of parallelrncorpus in order to achieve rnuency as in the case of statistical machine translationrn(SMT) and example based machine translation (EBMT). The contextrnawareness of phrase based machine translation (PBMT) approaches used forrnthe pair so far are also questionable. This research develops a system thatrntranslates English text to Amharic text using a combination of context basedrnmachine translation (CBMT) and a recurrent neural network machine translationrn(RNNMT). We built a bilingual dictionary for the CBMT system tornuse along with a target corpus. The RNNMT model has then been providedrnwith the output of the CBMT and a parallel corpus for training. The proposedrnapproach is evaluated using the New Testament Bible as a corpus. Thernresult shows that the combinational approach on English-Amharic languagernpair yields a performance improvement of 2.805 BLEU scores on average overrnbasic neural machine translation(NMT).

Get Full Work

Report copyright infringement or plagiarism