高激励项集的挖掘研究

基于支持度的关联规则只能找出所有的频繁集，无法找到那些非频繁但效用很高的项集；基于效用的关联规则致力于发现所有高效用项集，无法找到效用不高但支持度与效用的积很大的项集。为克服支持度与效用的不足，提出了一种新的项集重要性的度量方法（即激励）及一种自下而上的挖掘高激励项集的算法HM-Two-Phase-Miner。激励集成了支持度与效用的优点，能同时表达项集的语义特性与统计特性。HM-Two-Phase-Miner利用事务权重激励向下封闭特性进行减枝，有效提高了算法的性能。

Study of high motivation itemsets mining

Algorithms for support-based association rules mining can only discover frequent itemsets,but can not discover the non-frequent itemsets with high utility values;Utility-based association rules mining aims at discovering high utility itemsets,with-out considering the itemsets whose utility values are not high but the product of the support and utility of the same itemset is very large.To solve the problem,a new measure is proposed,i.e.,motivation,to measure the importance of an itemset and a down-top algorithm called HM-Two-Phase-Miner to discover high motivation itemsets.Motivation integrates the advantages of support and utility,and thus can reflect both the semantic significance and statistical significance of an itemset.In HM-Two-Phase-Miner algorithm,transaction-weighted motivation downward closure property is adopted to cut down the search space.