Research On The Algorithms Of Generating The Maximal Boolean Association Rules

Data mining which is the process of extracting the useful information and knowledge from the structured and unstructured datas has get the major progress and it has been applied in many fields. Association rule is one of the important research area in data mining. Its goal is to discover previously unknown, interesting relations from large databases.Firstly, this paper briefly introduces the development and the prospects of the data mining. We discovered that many reasearchers often focus on how to improve the efficiency of the algorithms of mining frequent itemset or the association rules. Actually, large of the association rules may be generated from the frequent itemset, and this problem may let the users didn't extract the useful information from the association rules efficiently. This paper focus on how to reduce the quantity of the association rules without lose information of association rules. The main reasearch is as follows:1.The algorithms such as GRSET which generate the association rules are thoroughly studied and they may lead to the"rules explosion".The quantity of the association rules which are get from frequent itemset may be reduced and the information of the association rules will not be lost. We present the GLMBR algorithm based on the research of the predecessor and this algorithm use the depth-first method generate the local maximal Boolean association rule which corresponding to frquent closed itemset and frequent key itemset. We analyzed the result of the GLMBR algorithm and the GRSET algorithm which generate assocation rule set of the same frequent itemset in theory and expriment which are used to proof the validity of the GLMBR algorithm.2.We present a algorithm of generating the MBR which based on the frequent closed itemset and frequent key itemset after we discovered the properties of the FCI and frequent key itemset,and the important relation between the maximal Boolean association rules and two special itemsets. Finally, we proof the algorithm's validity in theory and examples.