Association rule post-processing is a research challenge in KDD. In this post-processing task, objective interestingness measures are very useful for finding interesting rules possessing certain characteristics. Till now, the usual method for using objective interestingness measures is to select one or several suitable measures for filtering rules. This paper proposes a new approach to aggregate a set of interestingness measures using the Choquet integral as an advanced aggregation operator. Since an objective interestingness measure is considered as a point of view on rule quality, the aggregation of a set of objective interestingness measures can extract rules satisfying many points of view. The experiment is carried out on different groups (i.e. different natures) of objective interestingness measures to observe their behaviors.
展开▼
