- Machine Translation
- Post-Editing Machine Translation
- Managing projects involving MT
- MT Evaluation
Sessions on Machine Translation will include an overview of approaches to MT, including rule-based, but focusing on statistical MT, which is the basis of most of the currently implemented systems. State-of-the-art approaches including neural network models will also be presented. Participants will learn how to compile language resources for MT training and train their own domain-specific engines in several language pairs.
Post-Editing Machine Translation
Sessions on PEMT will include an overview of most widely used guidelines for light and full post-editing, including TAUS, Microsoft etc. Hands-on practice of post-editing strategies will be provided, focusing on how to maximize post-editing efficiency without compromising the quality of the output.
Managing projects involving MT
Projects involving MT bring changes to the workflow, pricing, human resources management and technical features. Various policies of integrating MT into professional translation environments will be discussed and critically evaluated.
Participants will learn about the basics of automatic and human evaluation of MT output. Standard automatic metrics such as BLEU, METEOR, TER etc. will be presented along with open-source software for their implementation. State-of-the-art human evaluation guidelines (ARPA, TAUS Dynamic Quality Framework) will also be conveyed, with the participants evaluating their own engines using different methods.
We would like to thank Memsource for granting us free access to their platform through the Memsource Academic Edition.
We would like to thank Crowdin for granting us free access to their localization platform through their academic license.