The massive amount of data, available from social media outlets, has spurred a vast variety of research. We have joined this stream of research in an attempt of identifying spatial patterns in certain topics, represented in social media communities. Inspired by our research on clustering methods for spatial data, we created this initial visualization for an initial visual analysis of the data, before diving deeper into the data.
Demo can be found here.
We are using a standard deviation ellipse on top to give a general sense of distribution for the tag, allowing to identify tags that are only used in a small geographic region vs. tags that are used in the whole area of interest (The code to calculate and draw a deviation ellipse, can be found here).
To highlight patterns beyond the overall distribution we are using the DBSCAN algorithm to highlight spatial clusters in the data. For the area of Berlin we can clearly see how many “touristic” tags are used in central Berlin as well as Potsdam and therefor form clusters in both areas.