Identifying the Most Influential User Preference from an Assorted Collection

A conventional skyline query requires no query point, and usually employs a MIN or MAX annotation only to prefer smaller or larger values on each dimension. A relative skyline query, in contrast, is issued with a combination of a query point and a set of preference annotations for all involved dimensions. Due to the relative dominance definition in a relative skyline query, there exist various such combinations which we call as user preferences. It is also often interesting to identify from an assorted user preference collection the most influential preference that leads to the largest relative skyline. We call such a problem the most influential preference query. In this paper we propose a complete set of techniques to solve such novel and useful problems within a uniform framework. We first formalize different preference annotations that can be imposed on a dimension by a relative skyline query user. We then propose an effective transformation to handle all these annotations in a uniform way. Based on the transformation, we adapt the well-established Branch-and-Bound Skyline (BBS) algorithm to process relative skyline queries with assorted user preferences. In order to process the most influential preference queries, we develop two aggregation R-tree based algorithms. We conduct extensive experiments on both real and synthetic datasets to evaluate our proposals