Representative skylines using threshold-based preference distributions

摘要: The study of skylines and their variants has received considerable attention in recent years. Skylines are essentially sets most interesting (undominated) tuples a database. However, since the skyline is often very large, much research effort been devoted to identifying smaller subset (say k) “representative skyline” points. Several different definitions representative have considered. Most these formulations intuitive that they try achieve some kind clustering “spread” over entire skyline, with k In this work, we take more principled approach defining objective. One our main contributions formulate problem displaying points such probability random user would click on one them maximized.