Question

I have a relatively large 2 mode network in adjacency matrix form of 4000 rows and 9000 columns I’m interested in looking at the relationships between the columns.

With a smaller network I would do t(matrix) %*% matrix, however I think a 9000^2 matrix would max out the memory on my computer.

One option is to convert the igraph adjacency matrix to the sna pacakge format, and use sna::gt, but I was wondering if there was similar functionality in igraph?

Thanks,

Answer 1

Use graph.incidence to construct the bipartite graph from your adjacency matrix, and then use bipartite.projection to project it to the corresponding one-mode network, based on either the rows or the columns.