Discovering block structure in networks

A generalisation of modularity, called block is defined. This a quality function which evaluates label assignment against an arbitrary pattern. Therefore, unlike standard modularity or its variants, network structures can be compared and optimal matrix determined. Some simple algorithms for optimising are described applied on networks with planted structure. In many cases the structure recovered. Cases where it not analysed found that strong degree-correlations explain so discovered pattern more ‘surprising’ than one under configuration model. well studied this new method, to automatically deconstruct in very useful way creating summary key features.