Re: [igraph] results of betweenness centrality and closeness centrality

[Gábor Csárdi]

There are a couple of (partially overlapping) usual reasons for the results being different:

- different software calculate slightly different measures, i.e. one normalizes the results, the other does not, etc.

- different software handle multiple edges differently

- they handle self-loops differently

- they have different defaults, e.g. some ignore edge directions by default.

Igraph uses the formulas that are in its manual, and the results are undefined (ie. I am not sure what happens) for graphs with loop and/or multiple edges.

Best,

Gabor

On Fri, Oct 4, 2013 at 9:16 PM, Jinie Pak <address@hidden> wrote:

I am currently using several SNA tools including R to analyze the data.  Eventually, I will use R because it can compute weighted network.
However, the results from R are different from other tools in terms of computing betweenness and closeness centralities.

In fact, the basic centrality scores from all tools are different except for the degree centrality.

I am just wondering what algorithm R uses for computing betweenness and closeness centrality.

I don’t expect R developers (experts) to know how other tools compute the centrality measures, but is there any possible explanation of why SNA tools produce different results?

I would like to know what causes this issue and any solution for this. 
Thank you! 

 

Jinie

_______________________________________________
igraph-help mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/igraph-help