[New Atlas Check] AbbreviatedAddressStreetCheck

[vladlemberg]

Description:

A check to validate that Address Street addr:street road type abbreviation is spelled out in full. Please reference official OSM good practice mapping conventions.

Corresponded Issues:
#597
#409

Potential Impact:

• Name Standardization.
• Improving Geocoding service.

Unit Test Approach:

Simulate unit tests covering most common address types and false positives.

Test Results:

Passed.

ISO	Total Flags	Sampled	Sampling %	TP	FP	False Positive Rate
USA	50,000	5960	12%	5912	48	0.8%

[Bentleysb]

This looks great. It looks like there is just a bit of configuration and sonar clean up needed. Should this only be enabled in English speaking countries, for the moment?

[atiannicelli]

This looks great @vladlemberg. Thanks! I agree with @Bentleysb that we should add a line to the configuration file that restricts this to a set of predominantly English speaking countries.

    "countries": [
      "USA", "GRB", "AUS", "CAN", "JAM", "IRL", <others?>
    ],

@Bentleysb - Do you think that having the street name txt file is the best way to store and access that information? It seems fine with me, but I know that we were curious if that would be acceptable, architecture wise.

[sonarqubecloud]

Kudos, SonarCloud Quality Gate passed!   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

96.3% Coverage
0.0% Duplication

[atiannicelli]

This looks great to me.

[Bentleysb]

Given we want to search on the values, would it make more sense to invert this map? It would be less memory efficient, as the road type would be duplicated for each abbreviation. However, it seems worth it for the increase in run time from being able to drop the double for loop in the flag method, given the scale of data this is expected to run on.