Fig. 2From: An illustrated approach to Soft Textual CartographyFor k=3 topics, Left: topic wordclouds (Fellows 2014) obtained from the C matrix for topic model with parameters: burning =4′000, 2′000 iteration, thin =500, seeds ={2′003,5,63,100′001,756}, nstart =5 and best =true. The colour scale is the inverse of the frequency of the word in the whole dataset (e.g. more specific words are darker than common words) whereas their size represents their importance in the topic. Right: related maps with the probabilities of the municipalities over each topic (Map base source: FSO)Back to article page