Top Subspace Synthesizing for Promotional Subspace Mining

Abstract

In Promotional Subspace Mining (PSM), Pat Ranking provides an effective way to locate the outstanding subspaces for a given object among a group of competitive objects in multidimensional data. In this paper, we propose a top subspace synthesis framework that aims to integrate the results produced through independent Pat Ranking procedures on multiple data sets. This solution framework provides a more sophisticated top subspace evaluation method, a potential solution that distributes the computation burden, and an effective strategy for mining promotional subspaces under streaming data setting.