Finding the most desirable skyline objects

Abstract

This paper introduces a new operator, namely the most desirable skyline object (MDSO) query, to identify manageable size of truly interesting skyline objects. Given a set of multi-dimensional objects and an integer k, a MDSO query retrieves the most preferable k skyline objects, based on the newly defined ranking criterion that considers, for each skyline object s, the number of objects dominated by s and their accumulated (potential) weight. We present the ranking criterion, formalize the MDSO query, and develop two algorithms for processing MDSO queries assuming that the dataset is indexed by a traditional data-partitioning index. Extensive experiments demonstrate the performance of the proposed algorithms. © Springer-Verlag Berlin Heidelberg 2010.