Approximate Continuous k Representative Skyline Queries over Memory Limitation-Based Streaming Data

Abstract

Continuous skyline queries over sliding windows is an important problem in the field of streaming data management. The query monitors the query window, and returns all skyline objects to the system whenever the window slides. However, this type of query is greatly influenced by the data set size, data dimensions, and distribution. In many cases, the scale of skyline objects may be large, leading that it is difficult for users to find suitable query results from a large number of skyline objects. In addition, the space cost may be very high. The state of the arts efforts cannot work under memory limited based environment.

To solve the above problems, in this paper, we propose a novel framework named $\rho$-AKRS(short for $\rho -$ Approximate K-Representative Skyline) to support k-representative skyline query under memory limited based streaming data. Unlike traditional k-representative skyline queries that retrieves query results based exact skyline objects, it selects k approximate skyline objects with high representative as query results. In order to support query processing, we propose a $\rho -$quad tree based index to support approximate skyline objects search, and propose a M-tree-based index to support k-representative skyline search. The comprehensive experiments on both real and synthetic data sets demonstrate the superiority of both efficiency and quality.