From documentary heritage to relational knowledge
The project brings together physical conservation and computational analysis to study historical newspaper collections and documentary heritage. Historical materials are stabilized, documented, and digitized before reproducible computational pipelines extract entities, normalize information, and connect actors, events, sources, and territories. These structured data are then modeled as relational and knowledge networks to reveal historical patterns, communities, and emerging narratives that are difficult to identify through individual document analysis.
We stabilize, document, and secure every artifact before digitization to protect provenance and material evidence.
Reproducible pipelines extract entities, normalize datasets, and assemble dynamic knowledge graphs.
Integrated methodology
Assess risks, prioritize pieces, and define capture strategies with detailed technical sheets.
Tag entities, events, and sources; harmonize formats; and map cross-collection correspondences.
Build concurrence/co-occurrence graphs to reveal historical patterns and emerging narratives.
Models & algorithms
Simulates preferential attachment to compare real networks with theoretical baselines.
Leiden/Louvain clustering segments sub-networks by period, actors, or territory.
Multilingual models cluster narratives and concepts for discovery tools and recommendations.
Colaboracion
We are onboarding municipal archives, family collections, and press fonds to expand the laboratory.
Email the lab