DOLFIN - Do cument- L evel Fin ancial test set for Machine Translation
Mariam Nakhlé, Marco Dinarelli, Raheel Qader, Emmanuelle Esperança-Rodier, Hervé Blanchon (1) Univ. Grenoble Alpes, CNRS, Grenoble INP , LIG, 38000 Grenoble, France(2) Lingua Custodia, 75008, Paris, France contact: [email protected] · 2025
Read DOLFIN - Do cument- L evel Fin ancial test set for Machine Translation by Mariam Nakhlé, Marco Dinarelli, Raheel Qader, Emmanuelle Esperança-Rodier, Hervé Blanchon (1) Univ. Grenoble Alpes, CNRS, Grenoble INP , LIG, 38000 Grenoble, France(2) Lingua Custodia, 75008, Paris, France contact: [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-03053.txt.
Description
Nakhlé and four co-authors build a test set for translating whole financial documents. Their 2025 paper notes that most test sets still work line by line. Finance texts shuffle facts across sections, so line tests miss real flaws. Their set, DOLFIN, groups text by section instead.
The set holds about nineteen hundred fifty aligned sections per language pair, across five pairs. The team lays out how the data was gathered and matched. Trials show the set can tell apart models that use context from those that do not. Weak spots in finance translation stand out clearly. The set is public for other groups to use.
Catalog record
| CREATOR | Mariam Nakhlé, Marco Dinarelli, Raheel Qader, Emmanuelle Esperança-Rodier, Hervé Blanchon (1) Univ. Grenoble Alpes, CNRS, Grenoble INP , LIG, 38000 Grenoble, France(2) Lingua Custodia, 75008, Paris, France contact: [email protected] |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-03053