Limitations of Skyline Algorithms
Saydiolim Ganiev, Aziz Nasridinov, Jeong‐Yong Byun · 2015
Skyline queries compute data that are not dominated by any other data in the same database. Thus, it can discover user preference points without using scoring functions. Until now a lot of algorithms have been proposed that can solve the given problem in different ways. Sort-Filter Skyline (SFS) algorithm first makes sorting on data, which is conducted based on a monotone function. And then it begins comparing data to find skylines. However, such method has some drawbacks. In this, paper we conduct investigation on how a monotone function affects on the algorithm results. We implement SFS algorithm, and run it on different datasets such as anti-correlated, correlated and uniform changing a monotone function such as sum and product each time. Experimental results show that the performance of SFS on different dataset is highly dependent on a monotone distance function. SFS with sum outperforms SFS used with product when it is applied on correlated and uniform dataset. Yet, on anti-correlated dataset SFS with sum performs inefficiently.