BOUQuET : dataset, Benchmark and Open initiative for Universal Quality Evaluation in Translation

ACL ARR 2025 May Submission649 Authors

Abstract: This paper presents BOUQuET, a multi-way, multicentric and multi-register/domain dataset and benchmark, and its broader collaborative extension initiative. This dataset is handcrafted in 8 non-English languages first, each of these source languages being represented among the most widely spoken ones and therefore having the potential to serve as pivot languages that will enable more accurate translations. The dataset is specially designed to avoid contamination and be multi-centric, so as to enforce representation of multilingual language features. In addition, the dataset goes beyond the sentence level, as it is organized in paragraphs of various lengths. Compared with related machine translation (MT) datasets, we show that BOUQuET has a broader representation of domains while simplifying the translation task for non-experts. Therefore, BOUQuET is specially suitable for the open initiative and call for translation participation that we are launching to extend it to a multi-way parallel corpus to any written language.

Paper Type: Long

Research Area: Resources and Evaluation

Research Area Keywords: Multi-way parallel dataset

Contribution Types: Data resources

Languages Studied: 55 languages

Submission Number: 649