Chuck D. wrote:
I decided to put together the USGS stuff, the maxmind free stuff
and the
GeoNames project files and in the end I had countries with no
states, states
with no cities and cities with no states. Some data sources said a
country
had 40 states, another said it had 50. It was difficult to try and
figure
out because I don't know geo stuff enough to verify it.
Yeah, all of our sources data has various degrees of noise. There
were even locations mis-typed as =countries= in the official NGA
downloads - you'd think their validation would at least identify
spurious countries :). We developed a set of heuristics for deciding
when two locations (usually but not always from two different
sources) were in fact the same entity. This was an area that needed
more work, however, when the project ended. In addition, different
sources had made different ontological decisions about what was
what. For instance, does the US have 50 states - what about the US
Virgin Islands, etc?
This was a few years ago - if we were to start up again, I suspect we
would investigate working with whoever is behind geonames.org, as
they seem to have the same kind of goals we did. Anyway, I will send
our schema under separate cover, and I will investigate sending you
the data as well.
- John D. Burger
MITRE
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
