Closeness centrality improvements

[Sébastien Heymann]

From http://forum.gephi.org/viewtopic.php?t=240&p=3931#p3931 :

I like that you used a normalized version of the closeness, however its independence of the graph is not so clear as we have discussed in our chapters 2-4 on centralities in Brandes and Erlebach (eds) of 'Network analysis -Methodological foundations', Springer Verlag, Heidelberg, 2005. I think that it is a bit confusing that the value which is computed here is called closeness centrality but actually is a normalized variant of it. What do you think of calling it differently? And it would be really helpful to get the raw values of the summed distance to all other nodes plus the number of nodes that are reachable by a given node. With these information, everyone could create their own normalization variant of the closeness. The closeness centrality itself is defined as the inverse of the sum of all distances from v to all other nodes w. It is not really defined (or rather: it is defined in too many ways) for disconnected graphs.