[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [igraph] a question about "similarity.invlogweighted()"

From: Tamás Nepusz
Subject: Re: [igraph] a question about "similarity.invlogweighted()"
Date: Tue, 24 Sep 2013 10:20:32 +0200

> Suppose I want to calculate the similarities between the nodes with ids
> {100,60,50,90}, if I set similarity.invlogweighted(graph,
> vids=c(100,60,50,90)). Then the fist row is for the node with id 100?

> I am trying to do it in the following way, do you think I am in a right
> way?
> (1) Firstly, give each node a number(from 0 to n-1) which is used as the
> id of the nodes, and record the ids for the m nodes.
You don't need that, igraph automatically assigns a numeric identifier to each 
node in the [0; n-1] range.

> (2) After I get a m*n matrix from similarity.invlogweighted(), I get the
> columns corresponding to the m nodes.

> So another question is that, if I input the graph with the function
> read.graph(file="graph_ids.txt",format="edge list"), will the numbers be
> used as the ids, or new ids will be assigned according to the
> sequence of the nodes? For example, if the first edge is "0, 100", will
> the node "100" be assigned an id "1"?
format="edgelist" uses the numeric IDs that you have in the file. The result is 
that if your IDs are not consecutive, you will have isolated nodes in the graph 
because igraph requires IDs to be consecutive in the [0; n-1] range. So, for 
instance, this edgelist file will give you a triangle:

0 1
1 2
0 2

But this edgelist file would give you a triangle plus 1000 isolated nodes 
because IDs from 0 to 999 (inclusive) were not used:

1000 1001
1001 1002
1000 1002

So, if your node IDs are consecutive from 0 to n-1, use format="edgelist". If 
your node IDs are not consecutive or you use non-numeric node IDs, then use 
format="ncol", which would assign IDs to the nodes from the range [0; n-1] and 
record the *original* identifiers from your file in the "name" vertex attribute 
(accessible via V(g)$name).

> BTW, would you please tell me where can I find the source code about
> similarity.invlogweighted()?
It is written in C and you can find it here:



reply via email to

[Prev in Thread] Current Thread [Next in Thread]