One common database or a network of databases with a common API?


#1

Is the goal of this project to create one common database, which every organisation can use to display their subset of data, or do we want each organisation to expose their existing database via a common API, to simplify exchanging data?

I learned about In Common at the Vinicia Mapmakers meeting, and based on the (brief) discussion and with our own project in mind I had the idea that we were talking about the second thing, a network of databases. But after reading some of the discussions here I think In Common wants to create one shared database, similar to how OSM is one database, but you can tag the nodes and filter on these tags to get the views you want. The main goal then would be to share resources, and keep costs down so everyone can focus on their added value to the world instead of recreating geographical databases.

For Vicinia I think the network idea would be more interesting, because we want to enrich the data, adapt it to our models, and it would be a weird thing trying to fit all of that in a database that should be suitable to everyone. There a Linked Data approach would be more suitable (imagine we can link each Repair Café in our database to a canonical reference in another database!), but the problem with Linked Data is that it is cool, but doesn’t seem to have much traction outside the academic world.


#2

Hi @jan.fabry, indeed Incommon seeks to create one common database, which every organization can use to display their subset of data, as well as data from other organizations. We want all organizations to provide data that can then be used elsewhere. Each organization can then tap in the common pool to display what they want, how they want it. The rest of your description matches how I see it as well.

I can’t see how this database could not be networked in the way you mean it: each partner indeed saves on infrastructure costs, but nothing prevents anyone from building richer data sets from there. One principle is that Incommon produces open data to encourage cooperation. Linked Data is certainly a way to present those common data sets and acquire more socio-geographical data from other sources. I think a long term view would be to implement WikiData-compatible APIs.