Tuning the cardinality of skyline

Skyline query has its own advantages and are useful in many multi-criteria decision support applications. But for the rigidity of skyline dominance relationship, the cardinality of skyline result cannot be controlled, either too big or too small to satisfy users' requirements. By relaxing the dominance relationship to k-skyline or general skyline, we propose a unified approach to find a given number of skyline. We call our output of skyline as δ-skyline, in which δ indicates the number of skyline result. Without any user interference such as assigned weights or scoring functions, we are the first to propose a method to tune the cardinality of skyline operator in both directions, to either increase or decrease according to the requirement of user. To tune the cardinality of skyline, we adopt the concept of k-dominate and also we propose a new concept of general skyline. A point p is in general skyline if p is skyline in some subspace. General skyline have their meaning for they are the best at some aspects and are good alternatives to fullspace skyline. Finally, we present two algorithms to compute δ-skyline. Extensive experiments are conducted to examine the effectiveness and efficiency of the proposed algorithms on both synthetic and real data sets.