Adapting Multilingual Embedding Models to Historical Luxembourgish

Andrianos Michail, Corina Julia Raclé, Juri Opitz, Simon Clematide · 2025

Read Adapting Multilingual Embedding Models to Historical Luxembourgish by Andrianos Michail, Corina Julia Raclé, Juri Opitz, Simon Clematide (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-2502-07938.txt.

Description

Michail, Raclé, Opitz, and Clematide adapt multilingual embedding models to historical Luxembourgish in a 2025 paper. Digitized old texts need good semantic search, but models struggle with scan errors and old spellings. Luxembourgish is a low-resource language.

The team collected old news articles and used a large model to split and translate sentences. This gave parallel training sets in German, French, and English. They also built a test set for cross-language search in old text. Existing models did poorly on that test. After training with contrastive methods or distillation, accuracy rose a lot for every model. Adapted models and data sets were released for reuse.

Catalog record

CREATORAndrianos Michail, Corina Julia Raclé, Juri Opitz, Simon Clematide
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-07938

Browse the full library · About the Tower of Babel