Mapzen Search: An open source geocoding api built on open data
mapzen.com
mapzen.com
MapZen makes a lot of great stuff actually. We also use the Mapzen Valhalla routing engine (https://github.com/valhalla) to do trail routing: http://blog.gaiagps.com/plot-a-precise-trail-with-the-gaia-g....
OpenStreetMap.org recently added Valhalla too - interesting to see competition between MapBox/OSRM and MapZen/Valhalla. The best mapping tech is being developed openly now, even competing open projects - that's a fount of innovation.
Working on a better one that highlights all the levers in the API :)
However, I found a few flaws after a couple of minutes of testing.
1. Neither search nor autocomplete appears to support filtering by fields other than country? We have an app that needs to filter autocompletions by state. A focus point might be sufficient if the autocompleter aggressively sorts results based on distance, but I have yet to meet one that does this (or allows a tuning of the distance function). A rectangular bounding box filter doesn't cut it. I guess I'll have to check out the code and see if this can be added/improved.
2. The API only returns points, with no boundary information, for street addresses. We have an app that uses geocoding to match street addresses to points. But if your address is a bit imprecise and matches an entire stretch of highway (it happens a lot!), that's obviously not a valid match — something like "Highway 66" is just wrong. Google will wisely return a bounds object that you can compute the area for and determine if it's small enough that it's likely a precise enough location (we use 10m2 as the maximum). Unlike Google, Mapzen doesn't even tell you that your match is a "route" — everything is an individual street address. Not sure how trivial this would be to implement, as boundaries would need support from datasets; OSM has the required data, OpenAddresses might not.
3. Street address matching seems a bit patchy. I tried my Brooklyn street address, and it didn't find it. The building was constructed last year, though, so maybe it's old data. Can someone from Mapzen comment?
1.) We do have a focus point option for autocomplete, with either a soft (will sometimes show you results far away if we think they're really relevant) or a hard (only stuff nearby) radius that can be set. Look at the boundary.circle and focus.point parameters in our /v1/search endpoint docs (https://github.com/pelias/pelias-doc/blob/master/search.md). As you probably realized tuning this for general use is actually quite difficult, but if you're running pelias yourself there's nothing stopping you from tuning to your particular use case. But if you're running your own instance you can also just add a filter by state, since we already have a state field in all our documents.
2.) That's right, even for our datasets that have polygon data (only OSM and Quattroshapes), we just take the center point. This is something we want to add in the future, but it may take a few months at least, since we don't currently have much infrastructure for handling polygon data. We also have another awesome team at Mapzen that is working on a gazetteer project called Who's On First(http://whosonfirst.mapzen.com) that should really help by offering concordance between datasets.
3.) This one's interesting, we should be able to fix it right away because NYC data is generally quite good. Can you check if your building is in OpenStreetMap(http://openstreetmap.org) or OpenAddresses (http://openaddresses.io has a download for the USA, then find the file for NYC)?
(1) Would it be easy to extend to filter by polygon? We want focus + boundary. We never want results that aren't within the boundary, but this is impossible with other geocoders because they will sort arbitrarily even if you specify a focus. For example, if you only want Montana, typing "San Francisco" should not return the city of SF.
(2) Is there a possible workaround that's easier to implement? For example, if we can ignore routes (= long roads where no numbered addresses were matched) that's an acceptable compromise until you support returning bounds.
(3) OSM finds it. The map doesn't have a polygon for it, so it's likely captured through interpolation. It matches as "house". I can send you the address privately if you want. OpenAddresses doesn't have it.
2.) Not likely for full polygon data, although we are thinking about doing something with TIGER data which has street address ranges for the US. I'm not quite sure what you're trying to accomplish but you're welcome to drop by our gitter room (https://gitter.im/pelias/api) and perhaps we can figure something out.
3.) It's probably poorly tagged data in OSM then, but again, we can chat about it on gitter more if you'd like. We re-import data once a week, so if you wanted to make changes to OSM, they should start showing up by Monday.
I've tried using the curl example provided in https://github.com/pelias/pelias and appending api_key=$apikey to it but I just get an error back saying that this is not a valid key.
Has anyone found any documentation for how to use this API?
We're super flattered over all the attention and we're listening for any feedback.