University of Ottawa wins Latin NER task using Gemini and Claude LLMs

Researchers from the University of Ottawa have achieved top results in the EvaLatin 2026 Named Entity Recognition (NER) shared task for Classical Latin. By employing prompt engineering with large language models Gemini 2.5 Pro and Claude Sonnet 4.5, they demonstrated the effectiveness of cross-lingual transfer learning for underrepresented ancient languages. Their system secured first place in both coarse-grained and fine-grained NER subtasks, outperforming all other submissions across various evaluation metrics. AI

IMPACT Demonstrates LLM capabilities for low-resource languages, potentially accelerating digital humanities research.

RANK_REASON Academic paper detailing research results on a specific task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• Classical Latin
• Claude Sonnet 4.5
• EvaLatin 2026
• Gemini 2.5 Pro
• named-entity recognition
• University of Ottawa

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →