Knowledge Graphs and Artificial Intelligence for the Implementation of Cognitive Heritage Digital Twins

Source abstract

This paper explores the integration of Artificial Intelligence and semantic technologies to support the creation of intelligent Heritage Digital Twins (HDT), digital constructs capable of representing, interpreting, and reasoning over cultural data. The study focuses on transforming the often fragmented and unstructured documentation produced in cultural heritage into coherent Knowledge Graphs aligned with the CIDOC CRM family of ontologies, particularly CRMhs and RHDTO. Two complementary AI-assisted workflows are proposed: one for extracting and formalising structured knowledge from heritage science reports, and another for enhancing AI models through the integration of curated ontological knowledge. The experiments demonstrate how this synergy facilitates both the retrieval and the reuse of complex information, while ensuring interpretability and semantic consistency. Beyond technical efficacy, the paper also addresses the ethical implications of AI use in cultural heritage, with particular attention to transparency, bias mitigation, and meaningful representation of diverse narratives. The results highlight the importance of a reflexive and ethically grounded deployment of AI, where knowledge extraction and machine learning are guided by structured ontologies and human oversight, to ensure conceptual rigour and respect for cultural complexity.