Interpretability and Refinement of Clustering

摘要: The difficulty to validate clustering reliability hinders the adoption of in real-life applications. We propose: (a) a set symbolic representations interpret problem spaces and (b) CluReAL algorithm refine any result regardless used technique. Both approaches are grounded by recently published absolute cluster validity indices. Conducted experiments show how refinement improves performances wide variety scenarios builds more interpretable solutions, whereas shown offer explainable summaries contexts. Refinement interpretability both crucial reduce failure increase performance control operational awareness processes that depend on clustering.