Previous  |  Next


Jan 9

Brady Forrest

Brady Forrest

Urban Mapping Gives Us Free Neighborhoods

For those of us who dwell in cities neighborhoods are an incredibly useful convention that is usually only known by locals. They break a city up geographically or architecturally or socially. Until now there has been no freely available source of neighborhood data for geohackers. As of this morning Urban Mapping is providing free access to their neighborhood database via an API. This same data powers the neighborhood features on AskCity and appears to be powering Google Maps (Radar post).
us neighborhood maps
Urban Mapping's data is impressive (if you want to test it out go to AskCity, they have the best neighborhood maps). Their "US coverage touches more than 1,000 municipalities, representing more than 20,000 distinct neighborhoods". The map to the right shows that coverage. Urban Mapping also have neighborhood data for portions of Asia and Europe.

The APIs are well-documented and have sections to guide geo-noobies through the commonly-used terms. Neighborhoods can be retrieved by lat-long, city and postal code. My zipcode 98122 has the following neighborhoods associated with it in the Urban Mapping data store: Broadway, Stevens, Yesler Terrace, Madrona, Leschi, Minor, Central Area, First Hill, Downtown, Capitol Hill. This is a good list (though it is missing the odd name used to describe my specific neck of the woods, Squire Park).

Surprisingly, the API is SOAP, not the ever-popular REST. If you aren't very familar with SOAP Urban Mapping has put up code samples in PHP and Ruby for each call. To use the API you must register first. There is no limit to the number of calls one can make, however like any free service it can be withdrawn at any time.

What can we expect from this? I expect to see neighborhoods start to appear in more mashups. (I hope WalkScore creates a walkability leaderboard for the neighborhoods of each city.) It would be great if neighborhoods got added to Housing Maps as well.

Why would Urban Mapping do this? It literally puts them on the map. If my prediction above becomes true their logo will start to appear on mashups. Aside from increasing name recognition I think that releasing the free API was a really smart thing to do business-wise. Neighborhood data, though incredibly valuable is not very common. People don't expect it on a site. By releasing their data into the wild Urban Mapping may be creating a market, a market where sites "have to have" neighborhoods.

tags:   | comments: 11   | Sphere It


0 TrackBacks

TrackBack URL for this entry:

Swashbuckler   [01.09.08 07:27 AM]

Interesting that they chose to use SOAP instead of REST. Anyone know why?

brady   [01.09.08 08:12 AM]

I put in a question about it. I'll report back

Adrian Holovaty   [01.09.08 08:31 AM]

An API that answers the question "What neighborhood(s) contain point X?" is a great first step, but what'd be *really* useful is access to the underlying neighborhood polygons, so developers could draw the neighborhood boundaries on their maps and otherwise perform geospatial queries on that data.

Granted, Urban Mapping needs to make money *somehow*, so I don't blame them for charging for that. :-) It's expensive to collect, it's expensive to maintain, and it's painful to convert into geodata. I bring this up because I think it's important to point out that they're *not* releasing the geodata -- they're just releasing *access* to it. That's a huge distinction.

Chandu Thota   [01.09.08 08:55 AM]

My detailed analysis is here:
At a heigh-level, there are three major issues with the API:
1. No Geometry
2. Non-Standard Neighborhood IDs
3. Lack of wire-format choices
As Adrian says, this is a first good step in the right direction, but not terribly useful for now...

Josh Knauer   [01.09.08 09:55 AM]

Chandu and Adrian's comments are spot on. Further, this is information that should be in the public domain. An effort like openstreetmap is needed to collect all of the different versions of neighborhood definitions for a given region.

The other issue is that these kind of "one size fits all" definitions of neighborhoods do not reflect local knowledge. For example, here in Pittsburgh, PA we have neighborhoods as defined by County government, city government, local neighborhood organizations, real estate agents, etc. There is no one standard definition for a neighborhood and information systems are needed that help to capture this level of complexity. This is a problem that none of the commercial resellers of public data (like UrbanMapping) deal with at all.

The solution to this problem is complex and needs a solid and open information architecture that can deal with multiple assertions for the shape of a given neighborhood name (just as one example).


Michael Bauer   [01.09.08 09:57 AM]

In addition to the polygon data availability I think that associating user-generated content with the neighborhood data would be a second step towards becoming a Source of Record and establishing alternative monetization.

Andrew Turner   [01.09.08 10:13 AM]

There are numerous "open" systems for sharing polygon geodata, especially around neighborhoods: ShapeWiki and CarTiki to just name two. However, it's quite a huge undertaking and will happen eventually.

However, in the meantime I disagree with Chandu, I think while it's not the full-data release, it still is incredibly useful. Providing a free mechanism for developers to easily provide 'hood search is a real value.

Regarding the "wire-format", the actual technology is fairly unimportant. An addressable URL for each 'hood would be great, but once you make a simple wrapper around any API, the interface questions become moot.

What would also be interesting would be exposing the time-history of the neighborhood definitions.

Tom Carden   [01.09.08 10:52 AM]

Granted this is not release of the dataset, just access to it. That's definitely a precise description of what's offered.

That said, I would challenge Adrian and anyone else planning to draw neighbourhood boundaries on a map to think carefully about whether that's even possible. Brady's example lists 10 (or 11) neighbourhood names for one zipcode - this will not be a neat patchwork, but a mess of overlapping polygons depending on who you ask.

And to address Josh's comment, I believe this service *does* reflect local knowledge. As I understand it this part of Urban Mapping's business is to address that exact problem - a neighbourhood is in the eye of the beholder.

Josh Knauer   [01.09.08 11:19 AM]


I may not have been clear on the local knowledge point. In many places, there are many different versions of the shape of a neighborhood, depending on who you are talking to at a local level. We need an information architecture that can handle many shapes for the concept of a given neighborhood. I applaud that Urban Mapping is doing *something* but I'm also recognizing that we all have a ways to go.


Michal Migurski   [01.09.08 04:11 PM]

This is fantastic, already have a few uses in mind for this service. Too bad about the SOAP though - what an unfortunate way to piss away a good thing with a poor choice of wire format!

Darrin Clement   [01.10.08 09:34 AM]

Big correction - AskCity is not using Urban Mapping's data - that is Maponics' neighborhood data. And Maponics licenses the full geometry - polygon databases.

Post A Comment:

 (please be patient, comments may take awhile to post)

Remember Me?

Subscribe to this Site

Radar RSS feed