Naver Papago Wins Award at International Competition for Self-Developed AI Translation Evaluation Model

Ranked 1st in 'Document-Level Quality Evaluation'

[Asia Economy Reporter Jin-gyu Lee] Naver announced on the 25th that Papago's AI translation evaluation model 'PATQUEST', developed in-house, was recognized for its translation technology competitiveness at the international machine translation competition 'WMT20'.

WMT is a globally prestigious conference in the field of machine translation, along with IWSLT and WAT, boasting the largest scale and longest history among the three. Papago participated in the WMT20 quality evaluation category, ranking 1st in 'document-level quality evaluation' and 4th in the English-German translation section of the 'sentence-level direct evaluation'.

PATQUEST is an evaluation model that assesses translation quality and enhances the performance of translation models. To evaluate translation accuracy, a vast amount of data comparing and assessing the original text and the translated text is required. PATQUEST automated the process of generating artificial data and applying it to model training, improving the accuracy of translation quality evaluation and the efficiency of model training.

Shin Jung-hwi, leader of Papago, said, "Papago's in-house developed quality evaluation model was recognized at an international competition," adding, "We will focus on developing core technologies to advance translation quality as a platform leading translation technology."

This content was produced with the assistance of AI translation services.