Cultural heritage collections available as linked open data (LOD) may contain harmful stereotypes about people and cultures, for example, in outdated textual descriptions of objects. Galleries, libraries, archives, and museums (GLAM) have suggested various approaches to tackle potentially problematic content in digital collections. However, the domain expertise and discussions about words and phrases used in LOD-collections are scattered across different resources and detached from the collections themselves. In this paper, we capture domain expertise about English and Dutch contentious heritage terminology in a knowledge graph. Contentious terms in the resulting graph are then linked to entities from other LOD-resources used in the cultural domain and beyond, including Wikidata and WordNet. We make our design decisions explicit and report on the linking process. The developed knowledge graph makes expert knowledge interoperable, so it can be reused by the cultural heritage community and other LOD-developers to contribute to a more inclusive representation of cultural heritage on the Web.

, , ,
doi.org/10.5281/zenodo.7713157
Culturally aware AI
European Semantic Web Conference
Human-Centered Data Analytics

Nesterov, A., Hollink, L., van Erp, M., & van Ossenbruggen, J. (2023). A Knowledge Graph of contentious terminology for inclusive representation of cultural heritage. doi:10.5281/zenodo.7713157