Skyline join query processing over multiple relations

Abstract

Skyline query on multiple relations, known as skyline join query processing, attracts much attention recently. However, most of the existing algorithms perform skyline join just on two relations. In this paper, we propose an efficient algorithm Skyjog, which is applicable for skyline join on two or even more relations. Skyjog divides each relation into two or three partitions. Based on the proposed group division approach, tuples generated by several join combinations of these partitions definitely are skyline points. Skyjog only has to examine tuples of other join combinations. Thus, Skyjog achieves performance efficiency by avoiding much skyline computation. Experiments demonstrate that Skyjog has an outstanding performance on all datasets, and outperforms the state-of-the-art skyline join algorithms on both two relations and more than two relations.