Research On SKYLINE Preference Query Technology Over Incomplete Data

Skyline query is a classical preference query. Due to its capacity of retrievinginteresting points from multi-dimensional datasets according to multiple criteria,skyline query problem has been studied broadly. Almost all the studies assuming thatdata dimensions are all available. However in many real life applications, likedecision making, location-based services and item rating information and so on.Because of unit exception, privacy protection and other reasons, almost all thecollected real datasets are incomplete.Data sets with missing attribute values or missing tuples are called incompletedata set. Here we mainly consider the case of incomplete attribute values in a data set.The traditional data cleaning approaches like filling and repairing method, howevereither spend too much time, or the accuracy is not high enough. In skyline query,every attribute is decisive in determining whether one point can be dominated byother points, and if it is not dominated by any other points，it is a skyline point. Sotraditional data cleaning methods are not suitable for incomplete data skyline problemfor its big influence on the query result.Currently, skyline query research on incomplete data has not been attractedenough attention, especially for the case of the large scale data set, they are muchmore tend to be incomplete, therefore the research for skyline queries on incompletedata is meaningful.In this paper, we proposes the idea of incomplete data dominant degree, anddesign the EIDS algorithm for static incomplete data sets, which can greatly reducethe comparisons between attributes among different tuples, effectively improve theefficiency of the algorithm save execution time overhead. Then we extend thisproblem to dynamic incomplete data sets, considering for the dynamic data set, wepropose sISkyline algorithm based on traditional sliding windows model.