Tharindu Ranasinghe lecturer at the University of Wolverhampton and a member of the Research Group on Computational Linguistics (RGCL), affiliated with the Research Institute of Information and Language Processing (RIILP). He holds a PhD in Computer Science from the University of Wolverhampton, which he defended in 2021. As a PhD student, he worked on applying deep learning-based text similarity models for applications in translation technology. He is the lead developer in TransQuest; award winning translation quality estimation software in WMT 2020, which has been downloaded for more than 18,000 times.
Translation Quality Estimation – Past, Present and Future with Multi-word Units
Summary – Over the years machine translation (MT) has progressed well to provide excellent results. However, there can be challenging scenarios for MT such as multi-word units, where quality estimation (QE) can be used to get an idea of the reliability of the translation. The goal of QE is to evaluate the quality of a translation without having access to a reference translation. In this talk, I am going to talk about the importance of QE and then past and present approaches. Finally, I will be talking about future directions in QE with special mentions to multi-word units.