Why is that when I do an all-core analysis of a network, the cores contain isolates? Shouldn't each core contain vertices that are all connected to n vertices within the core itself? How to avoid this problem?
From the definition of k-core it can be seen that k-core need not to be connected. After extracting the core, you can compute weakly connected components to see if the core is connected or consists of several pieces. AndrejBack to top