On 06/10/2010 15:45, 'Dragon' Dave McKee wrote:
I think Wikipedia might be a solution to my problem, as well. If we start
from the assumption that only villages and small towns have parish councils,
then the Wikipedia lists of civil parishes could be used to cross reference
with the OS Gazetteer to generate a reverse geocodable list of village
names.
I'm not certain that Parishes cover 100% of the land area.
They don't. But they don't need to, at least for my purposes.
The OS gazetteer has a list of towns and cities, but not a list of
villages per se. Instead, village names are all included as "O" (for
"other") type entries, which also includes a whole slew of miscellaneous
named places such as farms, manor houses, etc as well as subdivisions of
towns and cities.
By cross-referencing with the list of civil parishes, it ought to be
possible to extract all the "other" entries that have a parish council.
It's reasonable to assume that these are villages. So we can now update
the OS Gazetteer database to have a new code, "V", which can be applied
to all of them.
Going back to my geocoding, then, I can look up the nearest entry in the
gazetteer which has either the existing "T" (for "town") code or my
newly applied "V" code. Either way, I get the most likely answer to
"what do people call this place?" for any given location.
This would, however, be a whole lot easier if it were not for the fact
that Wikipedia, while it may be a great encyclopaedia, is an absolutely
crap database. There's a separate list of parishes for every county, but
the list pages don't have any consistent format so screen-scraping them
is likely to be somewhat tedious :-(
Mark
--
http://mark.goodge.co.uk
_______________________________________________
Mailing list [email protected]
Archive, settings, or unsubscribe:
https://secure.mysociety.org/admin/lists/mailman/listinfo/developers-public
