Priority-Based Skyline Query Processing for Incomplete Data

Abstract

Over the years, several skyline query techniques have been introduced to handle incompleteness of data, the most recent of which has proposed to sort the points of a dataset into several distinct lists based on each dimension. The points would be accessed based on these lists in round robin fashion, and the points that haven’t been dominated by the end would compose the final skyline. The work is based on the assumption that relatively dominant points, if sorted, would be processed first, and even if the point wouldn’t be a skyline point, it would prune huge amount of data. However, that approach doesn’t take into consideration that the dominance of a point depends not only on the highest value of a given dimension, but also on the number of complete dimensions a point has. Hence, we propose a Priority-First Sort-Based Incomplete Data Skyline (PFSIDS) that utilizes a different indexing technique that allows optimization of access based on both number of complete dimensions a point has as well as sorting of the data.