Towards Multilingual Automatic Dialogue Evaluation
This work addresses the lack of multilingual data for dialogue evaluation, which is incremental as it builds on existing methods with data augmentation techniques.
The paper tackled the problem of developing multilingual automatic dialogue evaluation metrics by leveraging a multilingual pretrained LLM and augmenting English dialogue data with machine translation, finding that careful curation of translations using quality estimation metrics outperformed naive approaches.
The main limiting factor in the development of robust multilingual dialogue evaluation metrics is the lack of multilingual data and the limited availability of open sourced multilingual dialogue systems. In this work, we propose a workaround for this lack of data by leveraging a strong multilingual pretrained LLM and augmenting existing English dialogue data using Machine Translation. We empirically show that the naive approach of finetuning a pretrained multilingual encoder model with translated data is insufficient to outperform the strong baseline of finetuning a multilingual model with only source data. Instead, the best approach consists in the careful curation of translated data using MT Quality Estimation metrics, excluding low quality translations that hinder its performance.