Finding the influence set through skylines

Abstract

Given a set P of products, a set O of customers, and a product p ∈ P, a bichromatic reverse skyline query retrieves all the customers in O that do not find any other product in P to be absolutely better than p. More specifically, a customer o ∈ O is in the reverse skyline of p ∈ P if and only no other product in P better matches the preference of o on all dimensions. The only existing bichromatic reverse skyline algorithm, which we refer to as basic, is designed for uncertain data. This paper focuses on traditional datasets, where each object is a precise point. Since a precise point can be regarded as a special uncertain object, basic can still be applied. However, as precise data are inherently easier to handle than uncertain data, one should expect that basic can be further improved by taking advantage of the reduced problem complexity. Indeed, we observe several non-trivial heuristics that can optimize the access order to achieve stronger pruning power. Motivated by this, we propose a new algorithm called BRS, and prove that BRS never entails more I/Os than basic. Besides our theoretical analysis, we also perform extensive experiments to show that in practice BRS usually outperforms basic by a large factor. For example, when both P and O follow the anti-correlated distribution, BRS is faster than basic by an order of magnitude. Finally, we address a new variation of bichromatic reverse skyline search where the conventional definition of dynamic skylines no longer makes sense. Copyright 2009 ACM.