A novel computational paradigm for clustering complex database objects extracted from distributed database settings via well-understood OLAP technology is proposed and experimentally assessed in this paper. This paradigm conveys in the so-called ClustCube cubes, which define a novel multidimensional data cube model according to which (data) cubes store clustered complex database objects rather than conventional SQL-based aggregations. A major contribution of this research is represented by effective and efficient algorithms for computing ClustCube cubes that, surprisingly, are capable of reducing computational efforts significantly with respect to traditional approaches. Our analytical contribution is completed by a comprehensive assessment of proposed algorithms against both benchmark and real-life data sets, which clearly confirms the benefits deriving from our proposal.
