Enriching Digital Heritage with LLMs and LOD

Our Lorentz Center workshop aims to explore the transformative potential of combining LLMs and LOD to enrich cultural heritage metadata in ways that foster FAIR usage. The workshop also provides a forum for knowledge exchange amongst participants with experience in CH standards, LOD, LLMs and NER, and explicitly engages with the challenges presented by validation, bias, and the ethical processing of heritage data. Workshop activities will utilise exemplars of CH metadata from international (Europeana) and national (British Library) organisations. Participants will be encouraged to bring their own metadata examples, including from non-Anglophone contexts. CH metadata typically includes a mixture of structured data and unstructured textual descriptions, the latter often several sentences in length. In many cases the fields within the schema that contain named entities are not populated. The workshop will make the problem space of working with Named Entities more tractable by dividing it into three core processes: Recognition, Disambiguation, and Relations, each explored at a specific stage of the workshop. The organisers have decided to keep this tight focus due to the limited time available.