Efficient User Preferences-Based Top-k Skyline Using MapReduce

As an important variant of skyline query, top-k skyline can find the best k points as the final results. In order to generate the results meeting the needs of users for massive data, we propose an efficient user preferences-based top-k skyline combining partially and totally ordered domains in MapReduce, named \(P/T\_SKY\_MR\). The whole course contains two main phases, partially ordered domains processing and totally ordered domains processing. In partially ordered domains processing, we propose the binary encoding and a pruning strategy to present the precedence relationship about the partially ordered domains and different user preferences. Meanwhile, in totally ordered domains processing, for finding the final results, a defined ranking criterion is also proposed in order to reduce the calculation cost and minimize the response time. A large number of experiments show that our method is effective, flexible and scalable.