
290 Text Mining and Visualization: Case Studies Using Open-Source Tools
for example, we could have pruned our initial graph not only according to the frequency of
the corresponding tags, but also according to the edge weights (which are also based on a
frequency measure). That way, we would have ended with a more clustered and less densely
connected graph, with a higher enough diameter and average path length. We chose not to
do so mainly in order to fully demonstrate the capabilities of the Infomap algorithm even
in such a dense graph, since the community detection is definitely easier for a graph that
possesses already some clustering on its own. ...