Why should I use bipartiteSBM?¶
control directly the numbers of communities to infer for a bipartite network. There are 2 numbers that we can specify; i.e., \(K_a\) and \(K_b\), one for each node type.
conclude a different partition with a smaller description length (and a higher AMI on tested synthetic dataset).
converge to consistent partitions.
estimate the SBM parameters parsimoniously (without over-fitting or under-fitting).
However, there are also some disadvantages of this program:
It’s slower than minimize_blockmodel_dl.
It is not guaranteed to find the globally optimal partition.