Browsing Robust Clustering-Alternatives

In the last years, new clustering approaches utilizing the notion of multiple clusterings have gained attention. Two general directions — each with its individual benefits — are identifiable: (i) extraction of multiple alternative clustering solutions from one dataset and (ii) combination of multiple clusterings of a dataset into one robust consensussolution. In this paper, we propose a novel hybrid approach to generate and browse robust, alternative clustering results. Our hybrid approach is based on frequent-groupings as specialization of frequent-itemset mining. In this way, the different benefits of the existing directions are combined, offering new opportunities for knowledge extraction.