Data mining is an area in computer science that aims to analyze the rapidly increasing amounts of business, scientific, and engineering data for knowledge and other profitable uses. Association rules mining is an important branch of data mining, which is used to describe the implicit relationship of the attributes in the transactional databases and has become an quite active field in the research of data mining in recent years.Mining frequent itemsets is the most important task of mining association rules. In this paper, the problems of mining maximal frequent itemsets, which is a compact representation of frequent itemsets and how to reduce the quantity of the association rules without lose information of association rules are researched. The main research is as follows:1. The Boundary algorithm whose function is to find the maximal elements of a down-set of direct product of finite finite-chains is thoroughly studied. The Boundary is a depth-first search algorithm, and it can be used to mining the maximal frequent itemsets problem, which has the property of down-set of position lattice.2. A new depth-first search algorithm, called GMPV, which accurately displays the itemset based on position vector, has put forward to mine the maximal frequent itemsets. In GMPV, the transaction database were mapped to a Boolean matrix and superset checking and pruning method based on support also were used to increase the algorithm efficiency. The experiment results displayed that the algorithm is validity.3. Based on the analysis of the definition and property of opened frequent itemset, the opened frequent itemset mining algorithm was designed, which can use to generate the maximal Boolean association rules. According to the property of the maximal Boolean association rules, this paper predigested the generating algorithm of maximal Boolean association rules. |