Research map / Large language models / Language, safety and interpretability
OccGen: Selection of Real-world Multilingual Parallel Data Balanced in Gender within Occupations
Marta R. Costa-jussà, Christine Basta, Oriol Domingo, André Niyongabo Rubungo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Translate with feminine vs masculine subset into Russian (human eval, English-to-Russian) | machine translation · OccGen high-resource | gender accuracy: 0.81 -> 0.68 | Hurt |
| Feminine vs masculine subset, English-to-Arabic | machine translation · OccGen high-resource | gender accuracy: 0.78 -> 0.41 | Hurt |
| Feminine vs masculine subset, English-to-Spanish | machine translation · OccGen high-resource | gender accuracy: 0.71 -> 0.62 | Hurt |
| Feminine vs masculine subset for English, Arabic and translating to Russian | M2M_100, mBART50_m2m, Opus-MT · OccGen high-resource | BLEU | Hurt |
| Feminine vs masculine subset when translating from Russian | M2M_100, mBART50_m2m, Opus-MT · OccGen high-resource | BLEU | Improved |
| Feminine vs masculine subset in low-resource English-Swahili | M2M_100, mBART50_m2m, Opus-MT · OccGen low-resource | BLEU | Improved |
| Alignment and balancing steps applied to en-es-ru-ar | OccGen pipeline · OccGen high-resource | number of entities: 14635 -> 2436 | Hurt |
| Balancing step applied to en-es-ru-ar | OccGen pipeline · OccGen high-resource | number of occupations: 203 -> 59 | Hurt |
| Alignment applied to en-sw | OccGen pipeline · OccGen low-resource | number of entities: 1371 -> 1053 | Hurt |
| Choose M2M_100 versus mBART50_m2m | MT model · OccGen low-resource | BLEU | Improved |
Closest papers by meaning
- Are Models Biased on Text without Gender-related Language? (ICLR 2024 · similarity 0.86)
- BeyondGender: A Multifaceted Bilingual Dataset for Practical Sexism Detection (AAAI 2025 · similarity 0.86)
- StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs (NeurIPS 2026 · similarity 0.85)
- GlotCC: An Open Broad-Coverage CommonCrawl Corpus and Pipeline for Minority Languages (NeurIPS 2024 · similarity 0.85)
- Auto-Search and Refinement: An Automated Framework for Gender Bias Mitigation in Large Language Models (NeurIPS 2025 · similarity 0.84)
- More of the Same: Persistent Representational Harms Under Increased Representation (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
