METM19 presentation
Machine translation (MT) post-editing has been a common linguistic practice for decades, but in recent years it has become even more widespread with the development of neural MT (NMT).
This session offers practical information for both experienced post-editors and those who are only considering exploring post-editing as a new milestone in their professional development. It will provide recommendations on the best strategies to maximize your post-editing performance. The session will have a strong focus on NMT and how the post-editing task is affected by this type of system compared to the previously prevailing statistical and rule-based types.
The session will cover the following topics:
About the presenter
Anna Zaretskaya works as a Machine Translation Implementation Manager at TransPerfect, coordinating all processes involved in MT usage to allow everyone to leverage the technology in the most efficient way. She is in constant communication with TransPerfect’s linguistic experts’ community, providing support in MT projects, developing training materials and leading webinar sessions on the topic. She holds a PhD in translation technologies and translators’ user needs and is an author of multiple publications in academic and industry journals.
Getting to the next level in post-editing neural machine translation
Anna Zaretskaya, Barcelona, SpainMachine translation (MT) post-editing has been a common linguistic practice for decades, but in recent years it has become even more widespread with the development of neural MT (NMT).
This session offers practical information for both experienced post-editors and those who are only considering exploring post-editing as a new milestone in their professional development. It will provide recommendations on the best strategies to maximize your post-editing performance. The session will have a strong focus on NMT and how the post-editing task is affected by this type of system compared to the previously prevailing statistical and rule-based types.
The session will cover the following topics:
- How to approach the task of post-editing with the right mindset. Depending on the system type, the output quality and the types of errors to expect are different.
- How to choose the best MT system for each project, including the most common techniques for MT system evaluation.
- How to quickly decide whether the MT output is useful. This includes the most common flags of “useless” MT output.
- How to spot the most challenging errors that are common to NMT systems in different languages.
- Special shortcuts and settings for post-editing in your translation tool.
- Best practices for ensuring the right translation quality that meets client expectations.
About the presenter
Anna Zaretskaya works as a Machine Translation Implementation Manager at TransPerfect, coordinating all processes involved in MT usage to allow everyone to leverage the technology in the most efficient way. She is in constant communication with TransPerfect’s linguistic experts’ community, providing support in MT projects, developing training materials and leading webinar sessions on the topic. She holds a PhD in translation technologies and translators’ user needs and is an author of multiple publications in academic and industry journals.
