Research on model of association rules mining with added-newly measure criteria
Dexing Wang, Xuegang Hu, Xiaoping Liu, Hao Wang, Jun Guo · 2005
Nowadays, association rules mining from frequent itemsets is an important task of data mining, which should satisfy two conditions: support and confidence. However, there exist some problems in the strong association rules mining. Firstly, there are a great number of redundant association rules, then it is difficult for users to find interesting association rules in them. Secondly, we ignored the weights of attributes, neglected more important attributes, therefore we should introduce new measure criteria to association rules mining, which are weighted interestingness and cover. In the paper, we think that the weighted interestingness and cover of measure criteria should be added to association rules mining, thus the weighted interestingness makes it easy for users to mine interesting association rules, and the cover of association rules makes it simple for users to reduce the amount of association rules, then the efficiency and veracity of mining association rules are improved.