Art Worlds and Knowledge Graphs: Mobilizing Data-Driven Approaches and AI for Large-Scale Provenance Research

Provenance research is complicated by the dispersal of collections and archival records across institutions. This presentation proposes a data-driven approach to large-scale provenance research, developed through the BECACO project at Leiden University. It explores how digital methods can connect heterogeneous museum data, reveal previously invisible patterns of collecting, and enable research across institutional and national boundaries. We introduce tools that ingest museum data into a queryable CIDOC-CRM Knowledge Graph, use Large Language Models for natural-language queries, and identify potentially overlapping actors across collections through fuzzy matching. Conceptualizing provenance Knowledge Graphs as “Art Worlds” (Becker 2008 [1983]), we highlight the interdependence of actors in knowledge creation. These approaches can expand provenance research while improving access to heritage for descendant and source communities.