Clustering Explainability: A Systematic Literature Review

• Guilherme S. Oliveira UFV
• Fabrício A. Silva UFV

Resumo

The increasing application and complexity of machine learning models have intensified the need for their transparency and explainability. Although much of the research in Explainable Artificial Intelligence (XAI) has focused on supervised models, unsupervised learning contexts, such as clustering, also present an inherent need for explanation. Thus, this work presents a systematic literature review dedicated to the explainability and interpretability of clustering algorithms. The review identifies, categorizes, and analyzes existing strategies for the explainability of clustering algorithms. The systematic review identified 23 related studies published between 2000 and 2025. Based on the analysis of these selected studies, the work highlights the primary techniques employed, emphasizing the prevalence of decision tree-based methods and the growing interest in model-agnostic approaches. The study also examines the primary challenges in the field, including the absence of labels, the inherent limitations of tree-based explanations, and the difficulty in accurately describing clusters with complex structures. As a final contribution, the work discusses future research directions, highlighting the importance of developing new explainability techniques, improving validation methods, and creating more solutions that are independent of the clustering algorithms employed.