Query Profiler Versus Cache for Skyline Computation

Abstract

A skyline query is multi preference user query which generates the best objects from a multi attributed dataset. Skyline computation in an optimum time becomes a real challenge when the number of user preference are large and size of the dataset is also huge. When such a big data gets queried at large, response time optimization is possible through maintenance of the metadata about the pre-executed skyline queries. We have earlier proposed, a novel structure namely �Query Profiler� which preserves such metadata about the historical queries, raised against a dataset. Also as the dataset gets queried at large, the dimensions of user queries often overlap and queries get correlated. Such correlations in user queries and the availability of metadata about the earlier queries, combined together speed up the computation time and the optimization of the response time of the further skyline computation becomes possible. In this paper, we assert the efficacy of the Query Profiler by comparing its performance with the parallel techniques which utilize cache mechanism for optimization of the response time. We also present the experimental results which assert the efficacy of the proposed technique.