ArtKB: A Multimodal Art Knowledge Base for Cultural Heritage

This repository contains the data release of ARTKB, a domain-specific multimodal Knowledge Graph for the visual arts. The package includes: artkb-rdf.ttl — the complete RDF/Turtle dump of the ARTKB Knowledge Graph; artkb-images.json — the list of all digital images associated with the entities in ARTKB, including stable identifiers and remote URLs for download. ARTKB is built through a fully operational extraction and modelling pipeline that ingests visual-arts–related entities from Wikidata, aligns their classes and properties to the CACAO ontology. In addition to the symbolic RDF layer, ARTKB links each artwork to one or more digital images whose metadata are provided in the accompanying JSON file. These resources support multimodal applications such as hybrid retrieval, visual similarity search, metadata enrichment, and AI-assisted curation workflows. The graph has 1.4M entities : Class Label # Resources cacao:CACAO_0000022 Artist 76K cacao:CACAO_0000023 Physical Artefact 307K cacao:CACAO_0000020 Museum 2.5K crm:E57_Material Material 2.6K cacao:CACAO_0000036 Movement 1.1K cacao:CACAO_0000027 Genre 1K cacao:CACAO_0000003 Digital Artefact 147K and over 8M triple, most common properties are: Property Label # Triples crm:P92_brought_into_existence 357K crm:P43_has_dimension 259K crm:P45_consists_of 241K crm:P4_has_time-span 233K crm:P108i_was_produced_by 161K crm:P138i_has_representation 147K crm:P62_depicts 100K crm:P53_has_former_or_current_location 97K cacao:CACAO_0000017 genre 77K cacao:CACAO_0000035 movement 6K