@Edent I've noodled around with trying to write some code to turn all of OSM into a Foursquare-style venues database but haven't had the time.
My rough plan—a script that parses the entire OSM dataset and:
- filters out the features one wants based on a predicate (pubs, restaurants, public buildings, etc.)
- flattens ways/relations into points (calculate centroids)
- emits lat, long, label, OSM type+ID
Stick result in DB (SQLite+SpatiaLite appeals to me), index on coords, query N closest.
