Learning Graph Representations by Dendrograms
This work addresses the need for better quality assessment in hierarchical clustering for complex networks, but it appears incremental as it builds on existing agglomerative methods.
The paper tackles the problem of evaluating hierarchical graph clustering by introducing a novel metric based on graph reconstruction from dendrograms, leading to a class of reducible linkages that produce regular dendrograms through greedy agglomerative clustering.
Hierarchical graph clustering is a common technique to reveal the multi-scale structure of complex networks. We propose a novel metric for assessing the quality of a hierarchical clustering. This metric reflects the ability to reconstruct the graph from the dendrogram, which encodes the hierarchy. The optimal representation of the graph defines a class of reducible linkages leading to regular dendrograms by greedy agglomerative clustering.