Worldwide database of locations (cities)?

Question

It's the first time I post in this community, I hope my question belongs here.

I'm currently developing an app where a user can add a location to his post, just like Instagram does (or Facebook). To do so, the user could either manually search for a location, or agrees to share his location so the app finds out where he is (geo-loc).

In the backend, for each post, I'll store the location name, location coordinates and a location ID.

I obviously thought about Google Maps in the first place, I then discover that Facebook has a Places API too. They both provide the data I need.

1.I was wondering if there's another API somewhere that could fit my requirements and that would be a bit more open and less intrusive regarding user privacy. I don't need a big precision, I believe only cities would be fine. I would also like to not put any money into that for now, as the app is a side project and might never generate any revenue.

2.Related question but maybe not for this stack exchange community: would it make sense to host a database of all cities myself (if that ever exists), or use an external service ?

Thanks

Answer 1

I suggest to try Geonames, which is a well-recognized open database of geolocalized places worldwide. Geonames links the name of places (from hotel to cities) to coordinates (longitude & latitude, in WGS84) as well as other information when available (population, zip code, ...). You can have a look of webservices based on geonames here.

Among the webservices, the FindNearby might suit your need:

Example here : http://api.geonames.org/findNearbyJSON?lat=49.6842&lng=5.8143&radius=10&username=demo

However, I don't know how to filter to output only "big cities" (and not the smallest populated place). An option is to make a query using a certain radius and then filter by population > your_threshold.

As stated here, do not use username=demo in production but create an account. For your second, related question, have a look at the terms of use of Geonames or contact them to check if you can use their services directly or if you have to implement them on a dedicated server.

Answer 2

Why don't you use OSM's API?

A similar question was asked on gis: How to get a list of OSM relations (id numbers) of cities and countries only?

Answer 3

check (and collaborate!) at Data Packaged Core Datasets the specific datasets:

• https://github.com/datasets/population-city
• https://github.com/datasets/world-cities

Both are related to https://github.com/datasets/country-codes

PS: about OSM and its APIs, see also http://nominatim.openstreetmap.org/

---

Others: specialized datasets or building a dataset from a "BigData of Datasets"

• at Wikidata, good controlled data, any SparkQL query!
  For example, see this query of all cities of Brazil

• The BigData of OpenStreetMap is available by Overpass-turbo.
  For example extrating all cities of Belgique.

• All citi-codes of Brazil, http://datasets.ok.org.br/city-codes

Answer 4

Check out citypopulation.de, a well-curated website of all the world's big citites and agglomerations.

The guy who maintains the site (I think it is only one person, Thomas Brinkhoff - see the Terms of Use) has put a lot of thought into how to draw borders around "the cities", on how to define accurate city limits and also estimate the populations living within wider metropolitan areas.

I don't think there is an API but hte data is suitable for screen-scraping, or donwloading in bulk.

Actually this type of data often needs to be downloaded only once (or rarely at best), so there is no need for a fast responsive API that can scale up to thousands of concurrent users.

Answer 5

I've developed an online REST service called GeoDB Cities that wraps the GeoNames data (and also leverages WikiData for translations). It should give you exactly what you need.

1. You'll need to get the user's current GPS location (which you can do via the GeoLocation API supported by most modern browsers).

2. Now just make a call to GeoDB to get the nearest city to that location (with a certain minimum population to exclude small towns). You can also optionally sort by largest cities first (if your use-case considers larger cities as more relevant).

For example, let's say you get the user's location as +33.915007-118.404626. Call the GeoDB Free tier to find the closest cities with at least 100,000 people and within 50 miles of these coordinates, sorted by population descending (largest cities first). Like this:

GET http://geodb-free-service.wirefreethought.com/v1/geo/locations/+33.915007-118.404626/nearbyCities?minPopulation=100000&types=CITY&radius=50&sort=-population&offset=0&limit=1

Now just take the first result, which should be Los Angeles.

NOTES:

• Setting limit=1 ensures that only a single result is returned.
• Setting types=CITY ensures that counties are filtered out. (Otherwise, you'll get back both types.)

Learn more about this use-case here:

http://geodb-cities-api.wirefreethought.com/guides/getting-user-current-city

And experiment here with a live demo:

http://geodb-cities-api.wirefreethought.com/api-reference/find-nearby-cities-for-location-api

The free instance lets you do up to 1 request/sec. If you need more, you can upgrade to one of the Pro tier plans.

Good luck!

Answer 6

I've recently been using here.com using the R library hereRand it seems to provide all the functionality I need, mainly geocoding text addresses and route finding between locations. Their 'Freemium' model allows up to 250k transactions per month.

Another, much more complex process would be setting up you own PostGIS server and loading the shape files for the area/country you need. I've done this for the US using the Census Brueau's TIGER/Line shapefiles and it works well once up and running but it can be complex process, especially if you are new to GIS. Also, I have no idea how it would work for areas outside the US.