3 ms·

I have been considering looking at OSM as "yet another data source" for my maps, but overall I run into the highly abstract issue of comparison: how do I know t

I have been considering looking at OSM as "yet another data source" for my maps, but overall I run into the highly abstract issue of comparison: how do I know that two segments (one from say OSM, one from perhaps the county) reference the same stretch of street for comparison? Even the slightest nudge of a single coordinate point on the line would mean that they were not geometrically identical, so I would not know to compare their attributes. Names are almost laughably bad for this purpose, for having multiple forms ("N SECOND ST" vs "NORTH 2ND STREET") and for being terrifying vague (so many, many Main Streets). Worse yet, a given segment might be split at a point not an intersection (speed change occurs here, municipality change, directionality change, road type change, etc) so I cannot even rely on those. And it is entirely possible for different maps to include a driveway, splitting a street at that point, while another does not, leaving it continuous. Large-scale automated comparison of this kind of thing is challenging.

galimaufry 6y ago
> Names are almost laughably bad for this purpose, for having multiple forms ("N SECOND ST" vs "NORTH 2ND STREET") and for being terrifying vague (so many, many Main Streets). libpostal (open source, open data location parser) might help with this. Between using it to parse the name, and supplementing this with the city name (using reverse geocode with imprecise coordinates), you might be able to get something useful.
juliansimioni 6y ago
One of the under-appreciated advantages OSM has here is that they have a fairly well enforced convention of using fully expanded abbreviations in addresses and street names. It's _much_ easier for automated comparison to replace `street` with `st` than it is to guess which one of many possible expansions to replace `st` with (is it saint? street? sankt? strasse?). This isn't foolproof but much of the "authoritative" data out there has no such convention, making analysis much more dangerous.
- at_a_remove 6y ago
  Oh, cramming the names into a common format is something I have under my belt, but there are additional wrinkles and, worse yet, local bizarreness (aka "historical convention"). The names are the least of it, really, even if they are ... odd.
- jonas21 6y ago
  The official USPS abbreviation list has some pretty crazy ones: https://pe.usps.com/text/pub28/28apc_002.htm https://pe.usps.com/text/pub28/28apc_002.htm
jakecopp 6y ago
https://sharedstreets.io/ https://sharedstreets.io/ may help! > SharedStreets Referencing System is a global, non-proprietary system for describing streets. The Referencing System is the foundation of the SharedStreets toolkit and it is used to connect a wide range of street-linked data. The Referencing System connects street data from private companies and cities. Referenced data can then be transferred between different maps seamlessly. I think it can align maps that have similar streets that may not align perfectly coordinate wise.

galimaufry 6y ago

> Names are almost laughably bad for this purpose, for having multiple forms ("N SECOND ST" vs "NORTH 2ND STREET") and for being terrifying vague (so many, many Main Streets). libpostal (open source, open data location parser) might help with this. Between using it to parse the name, and supplementing this with the city name (using reverse geocode with imprecise coordinates), you might be able to get something useful.

juliansimioni 6y ago

One of the under-appreciated advantages OSM has here is that they have a fairly well enforced convention of using fully expanded abbreviations in addresses and street names. It's _much_ easier for automated comparison to replace `street` with `st` than it is to guess which one of many possible expansions to replace `st` with (is it saint? street? sankt? strasse?). This isn't foolproof but much of the "authoritative" data out there has no such convention, making analysis much more dangerous.

at_a_remove 6y ago
Oh, cramming the names into a common format is something I have under my belt, but there are additional wrinkles and, worse yet, local bizarreness (aka "historical convention"). The names are the least of it, really, even if they are ... odd.
jonas21 6y ago
The official USPS abbreviation list has some pretty crazy ones: https://pe.usps.com/text/pub28/28apc_002.htm https://pe.usps.com/text/pub28/28apc_002.htm

jakecopp 6y ago

https://sharedstreets.io/ https://sharedstreets.io/ may help! > SharedStreets Referencing System is a global, non-proprietary system for describing streets. The Referencing System is the foundation of the SharedStreets toolkit and it is used to connect a wide range of street-linked data. The Referencing System connects street data from private companies and cities. Referenced data can then be transferred between different maps seamlessly. I think it can align maps that have similar streets that may not align perfectly coordinate wise.