How to handle very large bimodal edge list in R igraph?
Is there a suggested method to work with very large networks in igraph? I have a bimodal edge list with > 1.5M rows.
It looks like this:
Person Entity
1006542 2074562
1006542 2024740
1009621 2053211
I want to convert this to a unimodal network so I can calculate node centrality metrics. I am trying to use igraph's bipartite projection function. My machine (x64 with 32GB RAM) can't handle it, and neither can my Amazon EC2 instance with 64GB of RAM.
Here's my code
library(igraph)
huge <- read.csv(file.choose(), header=T)
biModal <- graph.data.frame(huge, directed = FALSE)
V(biModal)$type <-V(biModal)$name %in% huge[,1] #I want the Person column
summary(biModal)
IGRAPH f9a412d UN-B 1040327 1535552 -- + attr: name (v/c), type (v/l)
unimode <- bipartite.projection(biModal, which=TRUE)
I realize the resulting matrix is huge and sparse. I have looked at both bigmemory and ff, but neither seem to work with igraph. I feel there is something I'm overlooking. I don't usually work with networks this large, but it seems like this would be a common problem.
链接地址: http://www.djcxy.com/p/31870.html上一篇: CMake：我如何依靠自定义目标的输出？