For each data set we computed a single network structure that maximizes the posterior probability.
We also computed posterior probabilities of edges for every pair of variables. That is, although the
presented methodology supports computation of the posterior probabilities of arbitrary subgraphs,
here we only consider directed edges. Here we show results for the data sets Parity1 and Alarm
only. For the Zoo data set we are not aware of any correct structure, and for the Parity2 data set the
results are similar to the results for the Parity1 data set.