Concept Dissimilarity Based Approach for Skyline Relaxation

In the recent years, skyline queries become one of the predominant and most frequently used queries among preference queries in the database system. Based on the concept of Pareto dominance, skyline queries attempt to incorporate and provide a flexible query operator that returns objects (skylines) which are not being dominated (in sense of Pareto) by other objects in all dimensions (attributes) of the database. However, this process leads to two scenarios: either (i) a huge number of skyline objects are returned which are less informative for the end-users or (ii) a small number of skyline objects are retrieved which could be insufficient to serve the user needs. In this paper we tackle the second problem and we propose a new efficient approach to relax the skyline and increase its size. The basic idea is to build the fuzzy formal concept lattice for dominated objects, i.e., no skyline objects based on minimal distance between each concept and the ideal concept (i.e., the ideal object from the set of dominated objects w.r.t the user query). The relaxed skyline is given by the set S-relax, formed by the union of skyline objects and the objects of the concept who has the minimal distance to the ideal concept and the size of Srelax is equal to k. ( k is the user defined parameter). Furthermore, we develop efficient algorithm to compute the relaxed skyline. A set of experiments is conducted to demonstrate the effectiveness and efficiency of the proposed approach.