Learning Graph Representations by Dendrograms

TL;DR

This paper introduces a new metric for hierarchical graph clustering quality, enabling better reconstruction of graphs from dendrograms, and proposes a class of reducible linkages for regular dendrograms via greedy agglomerative clustering.

Contribution

It presents a novel metric for hierarchical clustering quality and identifies reducible linkages for constructing regular dendrograms in graph clustering.

Findings

New metric effectively assesses hierarchical clustering quality

Identifies reducible linkages for regular dendrograms

Enables improved graph reconstruction from dendrograms

Abstract

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.