Currently extracting geographic features from Wikidata requires an inefficient filtering through the already overburdened WDQS or downloading and handling the enormous dumps.
During the 2026 hackathon I want to create a Toolforge tool to regularly extract geographical features from the shared storage dumps and allow to download them from the web in common geographic formats (I was thinking FlatGeobuf and GeoParquet but other formats could be good candidates as well).
The critera for inclusion of items in these dumps will be
- absence of any P585 or P376 statements
- presence of a P625 statement
- absence in the P625 statement of any P582 qualifier or presence with a value later than the current time
- absence of any P580, P571 or P1619 statement or presence of them with a value earlier than the current time
- absence of any P582, P576 or P3999 statement or presence of them with a value later than the current time
If I will find enough time I will update the initialization data pipeline of OSM-Wikidata Map Framework to use and test these geographic dumps.
