This is an attempt to build a framework for agglomerative clustering of graph vertices. At each level, the model assumes that the edges are generated by a stochastic block model. The framework should handle weighted and unweighted graphs with a number of probabuility distributions for the edges such as Possion, Binomial, Guassian, etc. The parameter estimation at each level may be fully Bayesian, or maximum likelihood.
The class definitions for edges, graphs, and probability distributions, along with the associated algorithms are in the inc folder. The app folder contains the source code for the executables. The scripts folder contains some helper scripts.