[postgis-users] Incorrect Address Normalization

[Paragon Corporation]

I have this problem ticketed already:
http://trac.osgeo.org/postgis/ticket/1118
 
In this particular case you'd be actually better off deleting those records
and then it would normalize that particular variant right.
 
The problem is that all those other Cams that tiger does put the Cam in the
pre type field would then be wrong.
 
As noted in the last item of ticket I'm working with Steve Woodbridge to
have an alternative normalizer which should handle these issues better and
we are in the heat of it at the moment so very much in flux.
I haven't tried adding this entry to the new standardizer table set but will
and see how it fairs. But it does behave the same right now for this
particular case with the entries I currently have.
 
 
If you are interested your can download the code from here and compile
http://pagc.svn.sourceforge.net/viewvc/pagc/branches/sew-refactor/postgresql
/?view=tar

It creates a PostgreSQL extension you can install with
CREATE EXTENSION address_standardizer;
 
 
Lots of examples in this diatribe of this ticket:
http://trac.osgeo.org/postgis/ticket/2260


We are still working out some kinks as to how tiger is wrapping it and I
need to add in a few more entries.

In postgis 2.1

CREATE EXTENSION postgis_tiger_geocoder;
CREATE EXTENSION address_standardizer;

Then you can run  

SELECT pagc_normalize_address(youraddress);

Instead of 
SELECT normalize_address(youraddress);

And the tables used are tiger.pagc_lex, tiger.pagc_gaz, tiger.pagc_rules
which you can add entries to.


The other alternative more invasive solution which Steve suggested and he
does for his geocoding work is to normalize the tiger dataset itself.
I've been avoiding that solution (though it would work well with cases like
this), just because its going to add more time to the data load process and
also take up a bit more disk space.  Neither of which is all that appealing
nor ones I've benchmarked.  It will also require a bit more work to add that
logic to the prep step (though I think I probably overestimated the effort
on that part).

Thanks,
Regina
http://www.postgis.us
http://postgis.net



________________________________

From: postgis-users-bounces at lists.osgeo.org
[mailto:postgis-users-bounces at lists.osgeo.org] On Behalf Of Shawn Peterson
Sent: Thursday, April 25, 2013 12:22 PM
To: PostGIS Users Discussion
Subject: [postgis-users] Incorrect Address Normalization


I was testing normalize_address on the following address:



100 El Camino Real, Menlo Park, CA 94025


but it gives incorrect results. I tried to add the following entry to
postgis-2.1.0SVN/extras/tiger_geocoder/tiger_2011/tables/lookup_tables_2011.
sql:

    FROM (VALUES
        ('EL CAMINO', 'El Cam'),
        ('CAM', 'Cam'),
        ('CAM.', 'Cam'),
...

and then run
postgis-2.1.0SVN/extras/tiger_geocoder/tiger_2011/create_geocode.sh. But it
does not work.

My postgreSQL is 9.2.4, and postGIS is 2.1.0SVN.


Any help is appreciated!

Thanks,
Shawn