Improving Retrieval and Expression of Iconographical and Iconological Semantic Statements: An Extension of the ICON Ontology

Domain-specific data face the long-standing challenge of expressing granular knowledge in a structured way, which is fundamental for fostering information retrieval in the context of domain-specific semantic digital libraries. In the art history field, the ICON ontology strives to face such a challenge by providing a model for describing complex iconographical and iconological interpretations of visual artworks with a high level of granularity. Nevertheless, such a detailed model has drawbacks when applied to extensive real-world data. The need for an extension of the ICON ontology emerged during the creation of 1) the Iconology Dataset, a manually curated dataset representing a selection of the interpretations by the art historian Erwin Panofsky, and 2) IICONGRAPH, a knowledge graph (KG) created by re-engineering the iconographic statements of Wikidata and ArCo. In this work, we present the ontology extension made after the experience of creating such diverse resources. The updates cover two distinct aspects. Whereas some features needed a more thorough description, other interventions sought to simplify the ontology to optimize queries. To this end, we present the motivation for such modeling that emerged from the creation of the Iconology Dataset and IICONGRAPH. We evaluate the newly added features through competency questions and in terms of their efficiency.