This is a review paper, whose goal is to significantly improve our understa
nding of the crucial role of attribute interaction in data mining. The main
contributions of this paper are as follows. Firstly, we show that the conc
ept of attribute interaction has a crucial role across different kinds of p
roblem in data mining, such as attribute construction, coping with small di
sjuncts, induction of first-order logic rules, detection of Simpson's parad
ox, and finding several types of interesting rules. Hence, a better underst
anding of attribute interaction can lead to a better understanding of the r
elationship between these kinds of problems, which are usually studied sepa
rately from each other. Secondly, we draw attention to the fact that most r
ule induction algorithms are based on a greedy search which does not cope w
ell with the problem of attribute interaction, and point out some alternati
ve kinds of rule discovery methods which tend to cope better with this prob
lem. Thirdly, we discussed several algorithms and methods for discovering i
nteresting knowledge that, implicitly or explicitly, are based on the conce
pt of attribute interaction.