http://www.bibsonomy.org/burst/user/statphys23/adjacencyBibSonomy BuRST Feed for /user/statphys23/adjacency2008-07-21T01:37:08+02:00http://www.bibsonomy.org/bibtex/214b3f3b14f4d98c29c1452b57b0ddfe5/statphys23statphys232007-06-20T10:16:09+02:00topic-11 algorithm weighted clustering adjacency statphys23 matrix <span style="color:#555555;">A.K. <a href="http://www.bibsonomy.org/author/Krueger">Krueger</a> </span><em>Abstract Book of the XXIII IUPAP International Conference on Statistical Physics, </em><em>Genova, Italy, </em><em>9-13 July2007. </em>Wed Jun 20 10:16:09 CEST 2007Genova, ItalyAbstract Book of the XXIII IUPAP International Conference on Statistical Physics9-13 JulyNetwork Clustering by Adjacency Matrix Reordering2007topic-11 algorithm weighted clustering adjacency statphys23 matrix We present a deterministic approach for clustering weighted networks, by sorting the adjacency matrix by a line difference metrics. This parametrized metrics takes into account the direct connection of nodes, the structural equivalence ($N_1$ -neighbourhood), and the similarity of the $N_2$ -neighbourhood. For each such 2-parameter pair, the resulting clustering is an ordered adjacency matrix plus the optimal positions where to cut into subspaces; the final clustering is the one with the overall highest modularity. Looking at networks as matrices enables new views; one outcome was this algorithm, it produces results that are good to interpret. However, due to the full metrics between all lines and the exhaustive search in parameter space, it is not very fast yet.