TRACE system improves trustworthy retrieval in historical archives
TRACE: Accountable Agentic Retrieval for Source Discovery in Digital Archives
Artificial Intelligence
Summary
Historical archives are hard to search because the texts are often damaged, come from many different sources, and it's important to know where information originally came from. The authors created TRACE, a tool that helps find and trace the exact sources of information in old documents without needing complex retraining. TRACE was tested on French historical records and worked better than other methods, especially when answers required combining facts from multiple sources. It is also affordable enough for libraries or organizations without expensive computer equipment.
What this means in practice
- •For digital archivists: Implement a retrieval system that accurately identifies and traces sources within large historical document collections, improving research transparency.
- •For cultural heritage institutions: Provide affordable and scalable access to trustworthy source discovery tools for digitized historical archives without requiring expensive hardware investments.
Authors
Donghan Bian, Marie Puren, Florian Cafiero
Abstract
Historical archives pose a difficult retrieval problem for retrievalaugmented generation systems: documents are OCR-degraded, heterogeneous across genres and sources, and require strong source traceability for scholarly and institutional use. We introduce TRACE, a training-free agentic retrieval framework designed for accountable source discovery over historical corpora. The system was developed in the context of DECIDON, an interdisciplinary project on the circulation of political discourse between parliamentary debates and the press during the French Third Republic, involving digitised historical collections and institutional use cases. The prototype is currently deployed internally within the project and accessible to 24 researchers across six partner institutions. We evaluate TRACE on HistoriQA-ThirdRepublic, a benchmark of 1,752 French historical questions over parliamentary debates and newspapers from 1887, with documents derived from Biblioth{è}que nationale de France digitised collections. TRACE achieves R@10 = 0.856 and MRR = 0.653, outperforming sparse, dense, graph-based, and agentic RAG baselines, with the largest gains on multi-hop and cross-corpus questions. At approximately $0.02 per question under the default hosted inference configuration, TRACE also remains economically feasible for heritage institutions, laboratories or companies that cannot rely on costly local GPU infrastructure. These results suggest that, for large digital libraries and archives, retrieval accountability and corpus-aware agent design can provide a practical alternative to heavier training-based or graph-construction approaches.