Exception-Tolerant Skyline Queries.

This paper presents an approach aimed at reducing the impact of exceptional points/outliers when computing skyline queries. The phenomenon that one wants to avoid is that noisy or suspect elements "hide" some more interesting answers just because they dominate them in the sense of Pareto. The approach we propose is based on the fuzzy notion of typicality and makes it possible to distinguish between genuinely interesting points and potential anomalies in the skyline obtained.