Reading Comprehension in Czech via Machine Translation and Cross-lingual Transfer

TL;DR

This paper demonstrates that cross-lingual transfer with models like XLM-RoBERTa enables effective Czech reading comprehension without manual Czech training data, using translated datasets and multilingual models.

Contribution

It introduces a method for building Czech reading comprehension systems using translated datasets and cross-lingual transfer, avoiding the need for manual Czech annotations.

Findings

XLM-RoBERTa trained on English performs nearly as well as models trained on translated Czech data.

Cross-lingual transfer achieves competitive results without Czech training data.

The approach is adaptable to any language with sufficient monolingual texts.

Abstract

Reading comprehension is a well studied task, with huge training datasets in English. This work focuses on building reading comprehension systems for Czech, without requiring any manually annotated Czech training data. First of all, we automatically translated SQuAD 1.1 and SQuAD 2.0 datasets to Czech to create training and development data, which we release at http://hdl.handle.net/11234/1-3249. We then trained and evaluated several BERT and XLM-RoBERTa baseline models. However, our main focus lies in cross-lingual transfer models. We report that a XLM-RoBERTa model trained on English data and evaluated on Czech achieves very competitive performance, only approximately 2 percent points worse than a~model trained on the translated Czech data. This result is extremely good, considering the fact that the model has not seen any Czech data during training. The cross-lingual transfer approach is very flexible and provides a reading comprehension in any language, for which we have enough monolingual raw texts.