Distributed Skyline Computation of Vertically Splitted Databases by Using MapReduce

Skyline query retrieve objects that are not dominated by another object. A result of a skyline query is relatively small, does not contain less important objects, and is useful for selecting an object. In this paper, we consider a method for computing skyline query in MapReduce framework, which is a de facto standard in big data analysis. Currently, we have to be aware of data disclosure. Therefore, we propose a distributed computation method, in which each computer uses only a projected database that is vertically splitted from an original database, for computing skyline query. Since one computer can see only projected values, sensitive information in a database can be localized in the proposed method in addition to the advantage of the efficiency of MapReduce. Extensive experiments demonstrate the efficiency of proposed algorithm for synthetic datasets.