R - 有向图，仅在双向有向时保留

Question

I have a dataset of people that "like" each other. I converted it to a graph.

g.likes <- graph_from_data_frame(dt.likes, directed = TRUE)

But I need to filter out the people that have liked each other. so only keep if A->B and B->A.

I don't seem to be able to find how I can do this.

Thank you in advance

IGRAPH 80b2ed9 DN-- 15060 2577550 -- 
+ attr: name (v/c), week (e/n)
+ edges from 80b2ed9 (vertex names):
[1] 4  ->75  217->68  217->72  217->127 221->68  221->72  221->127 217->9   258->89  258->4   217->69  258->293 309->193
[14] 309->268 323->396 274->396 274->193 381->36  381->396 381->4   381->17  381->101 381->193 515->490 262->396 480->527
[27] 451->421 451->484 451->301 262->407 262->480 262->96  262->217 262->490 262->314 262->28  262->473 262->193 262->281
[40] 262->642 262->172 262->409 262->582 262->289 262->558 262->303 262->280 262->627 262->635 262->138 262->364 262->565
[53] 262->550 262->543 262->535 262->609 262->411 262->574 262->566 262->102 262->618 262->581 262->408 262->419 262->584
[66] 262->89  262->467 262->594 262->580 262->226 262->575 262->472 262->569 262->557 262->532 262->525 262->445 262->382
[79] 262->540 262->511 262->529 262->66  262->486 262->510 262->516 262->48  262->503 262->504 262->454 262->488 262->506
[92] 262->495 262->416 262->497 262->494 262->499 262->496 262->492 262->493 262->484 262->392 262->336 262->485 262->204

Answer 1

There is a function for this: which_mutual

g.likes3 <- subgraph.edges(g.likes, eids = E(g.likes)[which_mutual(g.likes)], delete.vertices = F)
plot(g.likes3)

Answer 2

You can do this by altering the adjacency matrix of the graph (you can do it other ways too but this is straightforward).

# Load the igraph package
library(igraph)

# make a toy graph to demonstrate 
admat <- matrix(c(0,1,0,1,0,1,0,0,0), nrow=3)
g.likes <- graph_from_adjacency_matrix(admat, mode = 'directed')
plot(g.likes)

# if A -> B and B -> A that implies that the (i,j)th entry
# of the adjacency matrix is equal to the (j,i)th entry, 
# (assuming that edges are binary and not real-valued)
# running the code starting from here on your own matrix will
# do what you want.
admat2 <- as_adjacency_matrix(g.likes, sparse = FALSE)
admat2 <- admat2 * t(admat2)

# now convert back to a graph structure
g.likes2 <- graph_from_adjacency_matrix(admat2, mode = 'directed')
plot(g.likes2)