Custom-trained vs. Generic: A Comparative Study of Post-editing Performance

Home/Custom-trained vs. Generic: A Comparative Study of Post-editing Performance
We are monitoring the COVID-19 situation with concern for the well-being of all our LocWorld network. We plan to hold LocWorld42 Berlin and hope to see you there. Should your conference plans be impacted by travel restrictions, illness or health concerns, please rest assured that your registration fee will be fully refunded.
Custom-trained vs. Generic: A Comparative Study of Post-editing Performance
Track: TAUS | TS1 |   Intermediate |
Wednesday, June 12, 2019, 9:00am – 9:45am
Held in: Room F5-7
Presenter:
Alexander Gigga - lengoo GmbH
Host: Anne-Maj van der Meer

Neural machine translation (NMT) has been superseding all other machine translation (MT) approaches for a few years now. All major MT providers only scratch the surface of the technology’s potential. The true power of NMT lies in the possibilities to tailor the engines exactly to your needs with a relatively low amount of your own data. In this session we will share our experience with post-editing these highly specialized MTs compared to generic MTs in terms of quality and speed for a couple of enterprise cases.

Takeaways: Attendees will learn the difference between a custom NMT engine and a generic MT; how custom NMT engines are built and what the data requirements to train them are; and how much better post-editing of custom NMT engines works compared to generic MT.