Researchers have developed a pipeline to transform flat cultural heritage records into a FAIR Digital Object (FDO) compliant knowledge graph using CIDOC-CRM. This system employs a large language model to distinguish between metadata values that should be persistent identifiers (PIDs) and those that can remain literals. The pipeline successfully links 86% of metadata slots and resolves 58.5% of previously un-enriched values, while also merging cross-lingual forms. AI
IMPACT This research demonstrates how LLMs can automate the creation of structured, machine-actionable knowledge graphs from unstructured data, potentially improving data discoverability and interoperability in cultural heritage.
RANK_REASON The cluster contains an academic paper detailing a new methodology for knowledge graph construction.
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →