Maximal cliques summarization: Principles, problem classification, and algorithmic approaches

Abstract

Several algorithms are available for computing all the maximal cliques of real-world graphs, both in centralized and distributed settings. However, in many application contexts, the sheer number of maximal cliques and their significant overlap call for strategies to reduce their quantity, maintaining only the most “meaningful” ones. In this survey we introduce a novel taxonomic framework that classifies summarization problems along two key dimensions: summarization principles and problem classes. Our framework provides a unified perspective on seemingly unrelated problems, organizing systematically the highly scattered literature on this topic, revealing underlying connections that were not previously well understood, and identifying several open problems in this field.