Efficient column-oriented processing for mutual subspace skyline queries

A mutual skyline query will enable some new applications, such as marketing analysis, task allocation, and personalized matching. Algorithms for efficient processing of this query have been recently proposed in the literature. Those approaches use the R-tree indexes and apply a series of pruning criteria toward efficient processing. However, they are characterized by several limitations: (1) they cannot process different interests on attributes for skyline and reverse skyline, (2) they require a multidimensional index, which suffers from performance degradation, especially in high-dimensional space, and (3) they do not support vertically decomposed data that is a natural and intuitive choice for the parallel queries. To this end, we address aforementioned these problems and propose three efficient algorithms, i.e., index-based mutual subspace skyline, optimized index-based MSS, and parallel mutual subspace skyline, using the column-oriented processing that is more suitable for subspace and parallel skyline. Extensive experimental results show that our proposed algorithms are effective and efficient.