Geospatiality: The effect of topics on the presence of geolocation in English text data

The past years have shown us the immense potential that lies in digital text data. The AI tools that are trained on such data can be immensely helpful and the data contains a wealth of information that we can use to better understand our world. And there is a huge amount of text data that can be used for spatial applications as well – all that it takes is that the texts have a geotag or mention a place. But is that an option for all types of text content? In my case, for example, when I am writing about international politics or sports events, I mention places and locations more frequently than when I write about philosophy.

We were asking ourselves if we can see similar differences in Big Data.

Therefore, in this study we analyzed this “Geospatiality” of topics across texts from different platforms, from web forums to Q&A sites and news pages. The results confirm our intuition: On all platforms we analyzed, the topic of a text substantially influences how likely they are to contain a geographical locations. And this is mostly consistent across platforms. However, there are also some surprising exceptions.

The key message that I take away from this study is that while there is an immense potential in the spatial analysis of text data, we should be aware that the topic and type of the texts we analyze plays a huge role in what we are able to see in the data – and that this might influence the tools and decisions we derive from them.

Recommended citation: Mast, J., Lemoine-Rodriguez, R., Rittlinger, V., Geiß, C., Biewer, C., & Taubenböck, H. (2025). "Geospatiality: The effect of topics on the presence of geolocation in English text data." International Journal of Geographical Information Science.
Download Paper