LA SELVA BEACH, CA causes parser confusion about the *state*

[glind-cc]

The Input Address

'360 CAM AL BARRANCO, LA SELVA BEACH, CA, 95076'

(ex. 123 Main St. Chicago, Illinois)

Current Output

ERROR: Unable to tag this string because more than one area of the string has the same label

ORIGINAL STRING: 360 CAM AL BARRANCO, LA SELVA BEACH, CA, 95076
PARSED TOKENS: [('360', 'AddressNumber'), ('CAM', 'StreetName'), ('AL', 'StreetNamePostType'), ('BARRANCO,', 'PlaceName'), ('LA', 'StateName'), ('SELVA', 'PlaceName'), ('BEACH,', 'PlaceName'), ('CA,', 'StateName'), ('95076', 'ZipCode')]
UNCERTAIN LABEL: PlaceName

When this error is raised, it's likely that either (1) the string is not a valid person/corporation name or (2) some tokens were labeled incorrectly

Expected Output

360 - AddressNumber
CAM AL BARRANCO - StreetName
??. - StreetNamePostType
LA SELVA BEACH - PlaceName
CA - StateName

Examples

• Fail: '360 CAM AL BARRANCO, LA SELVA BEACH, CA, 95076'
• Fail: ''360 CAMINO AL BARRANCO, LA SELVA BEACH, CA, 95076'
• OK: "360 Camino al Barranco, Selva Beach, CA"

Additional context

This is clearly meant to be "360 Camino al Barranco, La Selva Beach, CA", which is here: https://maps.app.goo.gl/bobVwDJjJ6ke5dRFA.

It's throwing a repeated label error, because it's see the "LA" in La Selva Beach as a state

Editorial

• features look correct
• parsing looks correct

I think the problem is the TAGGER.tag(features), which is labeling this as a state.

As an alternative solution, consider giving 'commas' weight while training, or special case 'states' somehow.

Thanks for looking into this!