Nominatim, the well-know OSM search engine,
has recently received a major update of the algorithm that computes addresses
from the boundaries and places in OSM. This talk first introduces the new
algorithm, explains the background and how it translates to tagging.
We'll then take a step back and explore the larger
picture what the evolution of tagging scheme of OpenStreetMap means for the users
of the data.
Nominatim is the search engine that powers the search box on the main OSM site.
One of the fundamental steps of preparing OSM data for searching is the extraction
of information about their location, commonly known as their address. This is not
a simple task because OSM data is much less structured than traditional databases
and many users revert to using external data for this reason. However, the data
is there and should therefore be usable. In the course of the last year the
address algorithm of Nominatim received a major overhaul to improve how addresses
are generated from OpenStreetMap data.
The first part of the talk introduces the new algorithm and how it came to be.
We'll look into the current state of tagging of boundaries, the problem of
the urban/rural divide and the difficulties of country-specific mapping.
The second part of the talk deals with the more general question what the
evolving tagging schema of the OpenStreetMap database means for data users.
The free-form tagging is one of the big strengths of OpenStreetMap. But the
lack of rules does not necessarily have to mean lack of order. Using the
example of address extraction, I'd like to discuss the future of the tagging
schema from the point of view of a data user.