Mahsa Mohaghegh
Doctor of Philosophy, (Computer Engineering)
Study Completed: 2013
College of Sciences
Citation
Thesis Title
English-Persian phrase-based statistical machine translation: Enhanced models, search and training
Read article at Massey Research Online: 
Over 130 million Persian speakers face a serious language barrier. Ms Mohaghegh was motivated to undertake her research in order to develop a system that can help remove this barrier. She developed a hybrid machine translation system that translates between the English and Persian languages, a language pair that has low resources in so far as there is limited data available for comparison. Her novel translation algorithms have been shown to yield very high quality translations for this language pair, particularly with the use of an automatic post-editing approach that smoothes translation output. Evaluation of her machine’s translation output far exceeds that of popular commercial systems (such as Google Translate). Ms Mohaghegh’s research has also opened up significant opportunities in future work, such as investigating the use of similar approaches for other low-resource languages like Maori.
Supervisors
Professor Rezaul Hasan
Dr Tom Moir
Associate Professor Abdolhossein Sarrafzadeh