Is there an open source address-parser(locations) algorithm?

Question

I'm looking for a good address-parser that could parser any free texts and generate a tokenized output:

If you have something in C# would be helpful, but has no problem if in another language.

Example 1:

Free text entry: 23 Street n. 213 new york ny

Output: Street: 23 Street, Number: 213, City: New York, State: NY

Example 2:

Free text entry: new york

Output: City: New York

Example 3:

Free text entry: 12345

Output: ZIP: 12345

Example 4:

Free text entry: Auckley Fake Street New York New York

Output: Street: Auckley Fake Street, City: New York, State: New York

Consider Brazilian addresses patterns, or something like that. — Zanoni, Jun 29 '09 at 17:56
There are some suggestions here: http://stackoverflow.com/questions/16413/parse-usable-street-address-city-state-zip-from-a-string — James A. Rosen, Jun 29 '09 at 17:57

score 3 · Accepted Answer · answered Jun 29 '09 at 17:54

3

Here is a Perl solution that seems to address your address problem:

http://cpan.uwinnipeg.ca/htdocs/Geo-StreetAddress-US/Geo/StreetAddress/US.pm.html

And here is JGeoCoder, a Java solution :

http://docs.codehaus.org/display/JGEOCODER/Parser

answered Jun 29 '09 at 17:54

mandaleeka

6,427
1
25
34

score 2 · Answer 2 · answered Jun 29 '09 at 17:56

If you're willing to take advantage of remote services, you can try Google's Geocoding API. I think it provides the detail you require while accepting free form input.

Edit: Sorry, I didn't catch the essence of your question. The Geocoding API cannot specifically identify what was typed in (for example, whether it was a ZIP code, an address, etc...)

Is there an open source address-parser(locations) algorithm?

2 Answers2