Research On A High Efficient Algorithm Of Frequence Items Mining

Data mining has been researched and applied widely in the near few years. Data mining of association rules which is highly demanded in the area of commerce decision-making is one of the most important and fundamental problems in this area. Because the step of finding the frequent itemsets is the bottle-neck of the data mining of association rules, most of the research attention is focused on efficient methods of finding frequent itemsets. In this thesis, the author gives a new memory-structure: Horizontal-Vertical-Link and a high efficient algorithm of frequent itemsets mining: Sorted-Horizontal-Vertical-Link Mine algorithm.Firstly, the thesis introduces the basic principle of data mining, association rules mining and frequent-itemsets mining.Secondly, the thesis analyses the strongpoint and shortcoming of the existed frequent-itemsets mining. Taking measures from memory-structure and mining means, the thesis puts forward Horizontal-Vertical-Link memory-structure and Sorted-Horizontal-Vertical-Link Mine algorithm. The improvement is achieved by scanning the database just once, reducing recursion count and building item horizontal-vertical link within transactions.Thirdly, basing on a lot of data, the thesis makes performance analysis between Sorted-Horizontal-Vertical-Link Mine algorithm and H-Mine algorithm. Then it analyses the advantages and disadvantages of Sorted-Horizontal-Vertica-Link Mine algorithm.At the last, the thesis brings forward the design idea of Sorted-Horizontal-Vertical-Link partitioning Mine algorithm and looks forward to the future.