Classification of trading networks with combinatorial optimization

To understand the structure of a trade market, network analysts search for patterns in the link structure of the network with the help of so-called blockmodeling algorithms. We present a new combinatorial way to measure the deviation of a network from having a given pattern. We show that even though the computation of the deviation is NP-hard, it can be efficiently solved to optimality by a branch-and-cut algorithm, being up to 10,000 times faster than comparable methods from the literature. We report results on the structure of the German photo trade market and use them to verify hypotheses on this market from the literature.