Here is fixed to 10 to reduce the data sparsity.

<p>Because the network is relatively small, the largest degree we get is only 30. Therefore, the results in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0110184#pone-0110184-g006" target="_blank">Figure 6a</a> just demonstrate that when the degree is small, how the sentiments' correlations vary with node degrees. While regarding to <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0110184#pone-0110184-g006" target="_blank">Figure 6b</a>, the linear bin is used to get emotion sequences for nodes with clusterings within the same bin.