The paper discusses how the traditional method of association rule mining based on user-defi ned minimum support and confi dence can result in either too many or too few association rules. This could lead to valuable information being missed or redundant rules being generated, which is not practical and can be costly to implement. |