Figure 2

Bipartite graph of topics and keywords. The graph was obtained through a Non-negative Matrix Factorization (NMF) process. For each topic, we select the top 10 keywords. Nodes with red labels are topics nodes where the color corresponds to the number of books under this topic (colors reflects the size of the nodes with gradient between yellow to red, indicating smallest and largest, respectively), and the size is proportional to the median sales of books under the topic. Nodes with red labels or blue nodes without labels are the keywords. For example, under topic Sport we see keywords like “team”, “fan”, “play”, under topic Science and Humanities we can find keywords like “scientist”, “planet”, “explore”. We also see that the topic Sport has a moderate number of books and its sales of the topic is one of the best. For the topic Science and Humanities, it has more books than Sport, but the sales of the topic is lower than Sport