Un corpus d'erreurs de traduction
Guillaume Wisniewski, Anil Kumar Singh, Natalia Aleksandrovna Segal, Yvon, François · HAL (Le Centre pour la Communication Scientifique Directe) · 2013
More and more datasets of post-edited translations are being collected. These corpora have many applications, such as failure analysis of SMT systems and the development of quality estimation systems for SMT. This work presents a large corpus of post-edited translations that has beengathered during the ANR/TRACE project. Applications to the detection of frequent errors and to the analysis of the inter-rater agreement of hTER are also reported.