I am starting to work with Gazetteers and trying to find something that will actually "mine" geospatial locations from sets of text. Academically, this is a difficult task as texts are based on context, and as such, things like THE BRIDGE, would be very difficult to identify. The Bridge is the Golden Gate, FYI(See how my second sentence lets you know where I am).

The end result that I am looking for is the ability to feed text into a tool and output locations with XY's. (Potentially with some level of "confidence", which would be asking for a lot).

I am trying to build a tool to feed on twitter and produce geospatial outputs(To visualize political opinions for academic purposes).

I have some experience with Python. I know this is a very complex question looking for a very simple answer.But any pushes in the right direction would be very helpful. And although this seems like a text-mining/data mining question, I believe it is also a GIS question, because it is surrounded by the concept of context, the idea of assigning some absolute space to a relational thing, as well as there is a matter of spatial accuracy, and precision related to the idea of assigning absolute coordinates to things.

1 Answer 1


The process you are attempting is called Geographic Information Retrieval (GIR), there are whole academic conferences about it. When I worked on this (~10 yrs ago) I collected some literature that may help you get started, for example papers I tagged GIR. Once you have extracted place names (toponyms) you need to disambiguate the place, and finally geocode them.

Disambiguation is the process of deciding which of the 1000 (or so) possible Londons your text refers to, geocoding is converting the name into an actual location.

I would start with Jochen L. Leidner's thesis Toponym Resolution in Text: Annotation, Evaluation and Applications of Spatial Grounding of Place Names which provides a good introduction to the issues.

As for code I would look at the geography package which should be good enough to get you started.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.