In many real-life applications (e.g., groups recommendation, investments selection, detection of fire/crime-most dangerous places, expert teams selection, etc.), users do not only require analyzing individual objects but also groups of objects to make decisions. Skyline paradigm is one of the most popular and useful approach for multi-criteria data analysis and decision-making. It aims at identifying a set of skyline points that are not dominated in Pareto sense by any other point in a dataset. In the recent two decades, the skyline definition has been extended with different variants and the skyline computation problem for finding the skyline of a given dataset has been studied extensively. One important problem that has been surprisingly neglected to the large extent is the need to find groups of points that are not dominated by others as many real-world applications may require the selection of a group of points.
In this talk, we focus on extending the skyline model to the combinatorial context to deal with group skyline set (i.e., groups that are not dominated by any other group in a given dataset). The theoretic setting used is the one based on Soft Computing. We present the recent developments on group skyline both from the semantic and computational sides. We also tackle explicitly the important issue related to controlling the size of group skyline. Some open challenges on the topic are discussed as well.