Detecting Hierarchical and Diverse Structure in Complex Networks Using MDL.

The traditional community detection method can only find the tightly connected structure, but the structure of the real network is hierarchical and diverse. Based on the minimum Description Length (MDL) principle, proposed a model consisting of the core node and 4 basic structures to represent the complex network and put forward a formula to calculate the minimum description length of network. First, delete the core in the network and get the hierarchical division, then decompose and merge the non-core network, and get the final network partition until the description length is minimal. Experimental results on real network dataset show that this method has better network compression effect, can get the hierarchical and diversified characteristics of the network, and get more realistic network partition. [ABSTRACT FROM AUTHOR]