Researchers have developed TRACE, a novel agentic retrieval framework designed to improve source discovery in historical archives. This system addresses challenges like OCR degradation and the need for strong source traceability in digital collections. TRACE was developed for the DECIDON project, focusing on political discourse during the French Third Republic, and has been deployed for 24 researchers across six institutions. Evaluations on the HistoriQA-ThirdRepublic benchmark demonstrated TRACE's superior performance compared to various baseline retrieval methods, particularly for complex multi-hop and cross-corpus questions, while maintaining economic feasibility for heritage institutions. AI
影响 Enhances AI's ability to accurately retrieve and cite information from complex historical datasets, potentially improving research in digital humanities and archival science.
排序理由 This is a research paper describing a new framework and benchmark for AI-driven retrieval in historical archives. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXivLabs
- Biblioth uevoque nationale de France
- CatalyzeX
- DagsHub
- DECIDON
- Donghan Bian
- French Third Republic
- Gotit.pub
- HistoriQA-ThirdRepublic
- Hugging Face
- ScienceCast
- TRACE
AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →
