Place types are typically associated with larger aggregate geographies, like Census Blocks and SPZs. Those geographies are typed according to demographic densities (dwelling units and employees by type) and street intersection density. UrbanFootprint's typing algorithm examines these densities and selects an UrbanFootprint place type that most closely matches the underlying data we have for that geography.
For institutional place types, we rely on point datasets and parcel land use codes. For example, a Census Block will get typed as Open Space by intersecting data sets like CPAD.
Typing of parcel canvases uses a similar methodology but includes parcel land use data where available.