Malti-Bias represents work carried out specifically to measure and mitigate bias in Maltese BERT-based models - BERTu (monolingual model) and mBERTu (further pretrained mBERT).
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
The source sentences in these datasets were compiled from various corpora. Please refer to their respective sources and licenses for more detail.
| Corpus | Source | Licensing |
|---|---|---|
| FLORES-200 | github | CC-BY-SA 4.0 |
| CrowS-Pairs | github | CC-BY-SA 4.0 |
| SEAT (eng) | github | CC-BY-SA 4.0 |
| Korpus Malti v4.2 | MLRS | CC BY-NC-SA 4.0 |
| Gender wordlist | github | MIT |
| Neutral & Stereotype wordlist | github | N/A |
If you use this data in your work, please cite the following paper:
Melanie Galea and Claudia Borg. 2025. From Measurement to Mitigation: Exploring the Transferability of Debiasing Approaches to Gender Bias in Maltese Language Models. In Proceedings of the 6th Workshop on Gender Bias in Natural Language Processing (GeBNLP2025). Download Paper