MaxMind and GeoNames as a starting point
I have a dataset and want to ask it “How Global? How Local?” and I think there are some options.
Some big piles of data change or morph so slowly they are ripe for cloning or ripping — places, postal codes, plant names, discographies.
Maxmind has claimed the places pie and gives a pile of data away while making its money by matching IP addresses to the records of cities, countries and regions. On the whole, they work in the open and with quite open resources — a glimpse at the MaxMind GitHub repository shows 30 projects updated this year. The MaxMind data most suited to my task is the Free World Cities Database which is lean and a little neglected.
GeoNames follows are more crowd-sourced approach and its output is comprehensive and up-to-date. So comprehensive that there are tens of millions of namespaces for geographic features like hills and lakes. This is overkill for the task I have in mind and I will use the Cities with a population greater than 1000 dataset.
I can feed off this geographic feast for free through the handily coded datasets but what are they providing?
Comparison Table
If I want more namespaces with a smaller semantic target and region data then MaxMind. For me this is “good enough”.
For great alternate name support, live editing of locations and fuller information then GeoNames is the choice.
Other Resources
Geocompare: a comparison of public and commercial geolocation databases
http://www.caida.org/publications/papers/2011/geocompare-tr/geocompare-tr.pdf
GeodataSource
http://www.geodatasource.com/world-cities-database/free