IICONGRAPH addresses a decisive weakness in cultural heritage knowledge representation: the difficulty of encoding not only what an image depicts but what its motifs, symbols and compositional relations mean historically. Sartini’s intervention refines and extends ArCo and Wikidata through requirements derived from research questions that existing graphs could not answer. This reverses a common ontology-building logic. Instead of imposing a comprehensive schema in advance, the graph is re-engineered from interpretive failures, allowing iconographic and iconological inquiry to shape the data model. The resulting resource treats meaning as relational and stratified, distinguishing depicted elements from symbolic and cultural interpretation while maintaining FAIR documentation, reusable algorithms and transparent evaluation. Its contribution lies in making art-historical interpretation computationally addressable without collapsing it into flat labels. The work therefore bridges iconology and the Semantic Web through an epistemic architecture in which representation remains extensible, contestable and traceable—conditions essential for any machine-readable system that claims to preserve the density of visual culture.