Machine Translation Pre-training for Data-to-Text Generation -- A Case Study in Czech

TL;DR

This study explores the use of machine translation pre-training to improve data-to-text generation in Czech, demonstrating significant performance gains and robustness in low-resource and unseen data scenarios.

Contribution

It is the first to apply machine translation pre-training for data-to-text generation in Czech, showing notable improvements over baseline models.

Findings

Pre-training significantly improves model performance.

Enhanced robustness to unseen slot values.

Effective in low-data scenarios.

Abstract

While there is a large body of research studying deep learning methods for text generation from structured data, almost all of it focuses purely on English. In this paper, we study the effectiveness of machine translation based pre-training for data-to-text generation in non-English languages. Since the structured data is generally expressed in English, text generation into other languages involves elements of translation, transliteration and copying - elements already encoded in neural machine translation systems. Moreover, since data-to-text corpora are typically small, this task can benefit greatly from pre-training. Based on our experiments on Czech, a morphologically complex language, we find that pre-training lets us train end-to-end models with significantly improved performance, as judged by automatic metrics and human evaluation. We also show that this approach enjoys several desirable properties, including improved performance in low data scenarios and robustness to unseen slot values.