The Spatial Nearest Neighbor Skyline Queries

TL;DR

This paper introduces spatial nearest neighbor skyline queries, a new user preference query type in spatial databases that evaluates locations based on distances to nearest neighbors, avoiding complex attribute weighting.

Contribution

It proposes a novel query type and an efficient N2S2 algorithm for computing it, improving performance over traditional skyline query methods.

Findings

N2S2 algorithm outperforms branch and bound in efficiency

New query type simplifies user preference specification

Applicable in service recommendation and investment planning

Abstract

User preference queries are very important in spatial databases. With the help of these queries, one can found best location among points saved in database. In many situation users evaluate quality of a location with its distance from its nearest neighbor among a special set of points. There has been less attention about evaluating a location with its distance to nearest neighbors in spatial user preference queries. This problem has application in many domains such as service recommendation systems and investment planning. Related works in this field are based on top-k queries. The problem with top-k queries is that user must set weights for attributes and a function for aggregating them. This is hard for him in most cases. In this paper a new type of user preference queries called spatial nearest neighbor skyline queries will be introduced in which user has some sets of points as query parameters. For each point in database attributes are its distances to the nearest neighbors from each set of query points. By separating this query as a subset of dynamic skyline queries N2S2 algorithm is provided for computing it. This algorithm has good performance compared with the general branch and bound algorithm for skyline queries.