Parallel and progressive approaches for skyline query over probabilistic incomplete database

The advanced productivity of the modern society has created a wide range of similar commodities. However, the descriptions of commodities are always incomplete. Therefore, it is difficult for consumers to make choices. In the face of this problem, skyline query is a useful tool. However, the existing algorithms are unable to address incomplete probabilistic databases. In addition, it is necessary to wait for query completion to obtain even partial results. Furthermore, traditional skyline algorithms are usually serial. Thus, they cannot utilize multi-core processors effectively. Therefore, a parallel progressive skyline query algorithm for incomplete databases is imperative, which provides answers gradually and much faster. To address these problems, we design a new algorithm that uses multi-level grouping, pruning strategies, and pruning tuple transferring, which significantly decreases the computational costs. Experimental results demonstrate that the skyline results can be obtained in a short time. The parallel efficiency for an Octa-core processor reaches 90% on high-dimensional, large databases.