This work investigates the semantic of 61 commonly used interestingness measures in order to explore their common and distinct characteristics, by means of a two-way contingency table of a pair of variables; A and B. As the first step, a synthetic data of six probability variables; P(AB), P(AB̅), P(A̅B),P(A̅B̅), P(A̅) and P(B̅) and profile of measurements are generated based on P(A), P(B), and P(AB). The exploration will be done based on semantic relationship. Secondly, an extension is done to characterize among 61 interestingness measures. Thirdly, their similarity and dissimilarity among the measurments are investigated in terms of association and correlation points of view. Finally, the semantic hidden in the properties of each measure is revealed.
展开▼
