The GKL team participated in the November OGC Metadata Code Sprint. Our approach was to focus on the semantic data enrichment, including metadata mapping, using RML. We picked creating an up-to-date placenames knowledge graph for Australia as our example, using the placenames ontology and the 2017 FSDF placenames project as inspiration. We created RML mappings for each state from JSON and CSV gazetteer data, following rules set in place name ontology. Nayomi led the team in the practical and technical guidance of using pyrml and RML mapper. Alexis provided some of the leadership around placenames, with Nenad leading the team overall. We’re still generating some of the final results, but in two days did manage to lay the foundations by:
- Generating an RML mapping from GeoJSON and CSV files for each state (with files are available on our GitHub repository);
- Mapping metadata statements associated to gazetteer data sets for each state.
- Understanding some of the idiosyncrasies of pyrml rules and mappings, for example, RML mapping differences between CSV data and JSON data.
Lots of things still to do, including wrangling differences between GeoJSON and WKT formats; fixing mappings for gazetteer lines and areas, not just gazetteer points; incorporating different rules from other ontologies including Geonames, DCAT, GeoDCAT; developing a dashboard/website for Australian place names like FSDF placenames project; amongst many other challenges for future work.
