Domain-Specific Text Generation for Machine Translation won a Best Presentation Award at AMTA 2022, one of the biggest MT conferences.
ADAPT PhD Student Yasmin Moslem has recently won the Best Presentation Award for the Research Track at the 15th biennial conference of the Association for Machine Translation in the Americas (AMTA 2022), one of the top conferences in Machine Translation. The Award was granted for the presentation of the paper titled: ‘Domain-Specific Text Generation for Machine Translation’ which is co-authored by Dr. Rejwanul Haque, Prof. John Kelleher, and Prof. Andy Way.
The research work introduces a novel approach to domain adaptation of machine translation systems. It specifically addresses the scarcity of in-domain data, required to customize baseline models. The paper proposes methods that can generate huge amounts of synthetic bilingual in-domain data, using large language models. Those proposed techniques achieve marked improvements in the translation quality of domain-specific texts.
This work is supported by the Science Foundation Ireland Centre for Research Training in Digitally-Enhanced Reality (d-real), the ADAPT Centre for Digital Content Technology, and Microsoft Research. Yasmin is currently pursuing her PhD in Machine Translation at School of Computing, Dublin City University.
To find out more about this award-winning Machine Translation research paper, click here >
