Findings of the Covid-19 MLIA Machine Translation Task

TL;DR

This paper reports on the Covid-19 MLIA Machine Translation Task, highlighting community efforts to improve translation systems during the pandemic using multilingual models, transfer learning, and data cleaning.

Contribution

It presents the results of a collaborative MT evaluation focusing on Covid-19 related translation, comparing scenarios with and without external resources, and emphasizing data cleaning.

Findings

Multilingual models and transfer learning were most effective.

Data cleaning significantly improved translation quality.

External resources provided additional benefits in some cases.

Abstract

This work presents the results of the machine translation (MT) task from the Covid-19 MLIA @ Eval initiative, a community effort to improve the generation of MT systems focused on the current Covid-19 crisis. Nine teams took part in this event, which was divided in two rounds and involved seven different language pairs. Two different scenarios were considered: one in which only the provided data was allowed, and a second one in which the use of external resources was allowed. Overall, best approaches were based on multilingual models and transfer learning, with an emphasis on the importance of applying a cleaning process to the training data.