Continuous skyline tracking on update data streams

Abstract

We consider the continuous skyline computation in a new scenario named as update stream, where the First-In-First-Out rule (which is the basic character in sliding window model) does not hold, causing the existing algorithms inapplicable. An algorithm called BUSM is developed and analyzed, which can continuous compute the change of skyline caused by update messages; then a progressive algorithm GUSM based on a novel grid-indexed data structure is proposed, which makes use of the character that deletion and addition operation appears simultaneously in update streams, and represents the influence region by grids for early elimination. Analytical analysis and experimental evidences show the efficiency of proposed approaches. © Springer-Verlag Berlin Heidelberg 2007.