Automated Construction of FAIR Digital Object Knowledge Graphs from Flat Cultural Heritage Records

Automated Construction of FAIR Digital Object Knowledge Graphs from Flat Cultural Heritage Records Poster presented at CIKM '26, the 35th ACM International Conference on Information and Knowledge Management, Rome, Italy, 7-11 November 2026. The FDO framework requires metadata values to be persistent identifiers wherever they denote reusable entities, yet cultural heritage aggregators store most values as plain text. We present a pipeline that turns flat Europeana records into an FDO-compliant knowledge graph structured with CIDOC-CRM, using an LLM to enforce the PID/literal boundary before any entity linking and to route resolvable values to Getty AAT, Wikidata, VIAF and PeriodO. On 637 archaeological records from five providers, it resolves 58.5% of the values Europeana left as plain text and merges 176 surface forms into 33 shared entities, including cross-lingual alignments such as Athens/Αθήνα; 17 of the 33 are correct on manual review. Graph connectivity does not separate the approach from string matching. What distinguishes the FDO graph is that every node is typed and resolvable. Paper: 10.1145/3799682.3840068 Code: github.com/ZResearch/aFDO_CIKM Keywords: FAIR Digital Objects; cultural heritage metadata; knowledge graph construction; entity resolution; CIDOC-CRM; large language models; Europeana; CIKM 2026 Zeyd Boukhers, Lingxiao Kong, Xenophon Zabulis, Georgios Toubekis Licensed CC BY-SA 4.0.