The Research On Association Rules With Multiple Minimum Supports

Mining association rules is one of the important research branches in data mining. Traditional association rule mining algorithms set a single minimum support threshold for all the items or item sets. But in real applications, different items have different occurrence frequencies. So the different items require vary minimum supports threshold. As a new research field of association rules mining, mining association rules with multiple minimum supports is regarded as an important research for mining association rules in rare items.Furthermore, for many applications, it is difficult to find useful associations among data items at low or primitive concept levels. Associations discovered at high concept levels may represent common and realistic sense knowledge. Multilevel association rules are confronted with the problem of a single minimum support threshold too, and the problem is more serious, because multilevel association rules are extended association rules.This paper analyses and generalizes the property of mining association rules with multiple minimum supports, presents two new algorithms to discover multilevel association rules with minimum supports associated with the specialty of mining of multilevel association rules and based on the improvement on two algorithms. The two new algorithms resolve the question of rare items. Based on the theorems and their proofs, the two algorithms are optimized. Theories and examples prove their feasibility and efficiency.