Geoinformation

[m-appel]

Collecting different sources for geoinformation here that could be helpful with modelling:

• Geonames
• Geofeed

Geofeed only contains Country and City name, which might break thinks in the US where it is not uncommon to have multiple cities with the same name in different states.

In general normalization (regions, city names, etc.) between different datasets might be problematic.

[romain-fontugne]

For Geonames we'll have geo-coordinates (lat,lon) that we need to model.

I suggest to start modelling the coordinates that are already present in the database. I found coordinates related to AtlasProbe, AS (data from ASRank), Facility and IXP from PeeringDB.

For these entities, we can add a relationship LOCATED_AT to a new node type Point that gives their geographical position.
Something like this:

(:AtlasProbe)-[:LOCATED_AT]->(:Point {position: point({longitude: 12.3, latitude: 45.6})

Then, in a second step we can add geonames cities with their coordinates. This would allow to do most of the basic searches, e.g. find all resources around one city.

Finally, if we want to map better Points to their regions/cities we can import geonames admin areas and corresponding polygons, and do a one-to-one mapping in a post-script.