GeoNames Forum

Hi,

I was wondering how you actually extract the coordinates from Wikipedia, as there are so many different and ever-changing templates for specifying them. Do you use regular expressions? If so, are they available somewhere?

This would be great, because we are also extracting coordinates from Wikipedia, but you seem to have a higher recall (according to your data, you have around 90 000 entities from the English Wikipedia alone, we only have 60 000).

This is actually part of a project of merging Wikipedia and Geonames to a structured knowledge base which will also be freely available under Creative Commons, so we won't steal anything

Thanks in advance!

parsing wikipedia is a pain, sometimes it is better sometimes it is worse it depends on what kind of robots are just messing around when you want to parse it.

well, creative commons has many licenses. Not all of them can be considered free. As you are working with wikipedia you are likely forced to use the share alike type, which cannot be considered free.

Marc

Hi Marc,

thanks for the quick reply. Regarding licenses - you are working with Wikipedia data but still use the Attribution license, right? Same as us

Cheers

No, the GeoNames dataset does not include wikipedia data. We are just linking to it not merging wikipedia with our content.

Marc

Ah, i see, so you have your own coordinates and just parse Wikipedia to match them by geographical vicinity?

Sorry for the misunderstanding - so is there any possibility to get to know how exactly you parse the coordinates?

there is no secret trick I could possibly post here and everything is fine. It is just a big mess for everybody trying to parse it. I could spend a day looking at the source code and post a summary, but I would prefer using the time to refactor the code instead.

Marc