Dr Frédéric Blain, University of Wolverhampton
Shared tasks in NLP
11 November 2020
Shared tasks have an important role of identifying interests for complex problems in a field, as well as to quantify progress made during a given period of time. In this seminar, we will revisit the History and key aspects of shared tasks in the field of Natural Language Processing. I will share with you my experience as co-organiser of the Quality Estimation Shared task at the Conference on Machine Translation (WMT). Finally, we will discuss some ethical considerations that are arising in the field with regard to the organisation and participation to such challenges.
Fred Blain is a Senior Lecturer of Translation Technology at the University of Wolverhampton and a member of the Research Group on Computational Linguistics (RGCL).
Prior to joining RGCL, Fred was a research associate in Machine Translation in Prof. Lucia Specia’s group at the University of Sheffield. There he worked on discriminative training algorithms for Statistical Machine Translation and continuous adaptation from post-editing workflow within the scope of the EU H2020 QT21 project. He then turned to Quality Estimation for Machine Translation, a topic he has been working on since, in close collaboration with Lucia Specia. Together, they successfully secured several research grants (an Amazon Research Award grant, an EAMT grant and more recently Bergamot, an EU H2020 project), leading to many publications and DeepQuest, the first open-source toolkit for quality estimation for neural-based Machine Translation.
Fred holds a PhD in Computer Science from Le Mans Université (France), which he defended in 2013. As a PhD student, he worked on post-editing, continuous adaptation as well as domain and project adaptations for Machine Translation under the supervision of Holger Schwenk (Facebook) and Jean Senellart (Systran). He pursued his PhD work as a postdoctoral researcher at LIUM, the Computer Science Laboratory of Le Mans Université, by joining the EU FP7 MateCAT project. He also has experience in industry having held a research engineer position at Systran during his PhD.