An enhanced a priori algorithm for mining multidimensional association rules

Two concepts from different research areas are brought together, namely functional dependencies which are a class of integrity constraints that have gained fundamental importance for relational database design and association rules which are a class of patterns, which has been studied rigorously in data mining. It is shown that functional dependencies may be used to logically infer new association rules from given ones. This observation will then be employed to propose a new variant of the best known algorithm for association rule mining, the so-called a priori algorithm.