Getting Critical Categories of a Data Set

Ranking query that is widely used in various applications is a fundamental kind of queries in the database management field. Although most of the existing work on ranking query focuses on getting top-k high-score tuples from a data set, this paper focuses on getting top-k critical categories from a data set, where each category is a data item in the nominal attribute or a combination of data items from more than one nominal attribute. To describe each category precisely, we use a data distribution that comes from the score attribute to represent each category, so that the set consisting of all categories can be treated as a probabilistic data set. In this paper, we devise a novel method to handle this issue. Analysis in theorem and experimental results show the effectiveness and efficiency of the proposed method.