Figure 4
From: Stability of similarity measurements for bipartite networks
The μ–σ location map for (a) reshuffled MovieLens data set with parameter η = 0.5 and (b) randomly generated data set with M = N = 5000 and 〈ko〉 = 20. On each map, we still mark the clusters we observed in the empirical results which are shown in Fig. 3. It turns out that, measurements in the same clusters still have similar similarity stabilities even when the user behavior patterns are removed. Thus, the similarity stability patterns are due to the nature of each measurement, not the properties of the data sets.
