Domination Mining and Querying

Abstract. Pareto dominance plays an important role in diverse appli-cation domains such as economics and e-commerce, and it is widely be-ing used in multicriteria decision making. In these cases, objectives are usually contradictory and therefore it is not straightforward to provide a set of items that are the “best ” according to the user’s preferences. Skyline queries have been extensively used to recommend the most dom-inant items. However, in some cases skyline items are either too few, or too many, causing problems in selecting the prevailing ones. The num-ber of skyline items depend heavily on both the data distribution, the data population and the dimensionality of the data set. In this work, we provide a dominance-based analysis and querying scheme that aims at alleviating the skyline cardinality problem, trying to introduce ranking on the items. The proposed scheme can be used either as a mining or as a querying tool, helping the user in selecting the mostly preferred items. Performance evaluation based on different distributions, populations and dimensionalities show the effectiveness of the proposed scheme.