Crowdsourcing as a Method for Producing MT Training Data

Crowdsourcing as a Method for Producing MT Training Data

Track: TAUS | TS8 |
Friday, June 10, 2016, 11:30pm – 12:15am
Held in: Liffey Hall 2
Presenter:
Selçuk Özcan - Transistent
Host: Jaap van der Meer

This session is intended to present a crowdsourcing case study inspired by a project for which the main goal was to obtain a training data set. A very rare language pair, quality expectations, a huge volume of content and a limited time frame… How can one deal with these conditions? Briefly, it will be a great opportunity to discuss the methodology and the tactics to be adopted in this very specific scenario.