The Research Of Bitwise-based Closed Pattern Mining Algorithm

Nowadays, with highly developed information technology in society, a large amount of historical data has accumulated in the real-life and commercial applications, and these data are in highly explosive growth. Vast amounts of historical data do not only contain a large number of valuable resources, but also make us submerged in the huge wave of data and information. In order to find out potential and useful knowledge from the massive amounts of data, Data Mining Technology has emerged and shows strong vitality and huge potentiality. Frequent pattern mining has always been playing an important role in data mining tasks. Frequent pattern mining is a time-consuming process, and may produce a large number of frequent patterns. There is a frequent closed pattern which contains the same information with less quantity in outcome than frequent pattern.Frequent item sets mining is the first step in the generation of the association rules, so its mining efficiency is directly related to the efficiency of generation of association rules. This paper will study the two-dimensional and three-dimensional closed frequent item sets with a bitwise processing technology. In order to make the best of the computer processing 32-bit data each time, the data sets and the item sets will be stored by bits, so that it can increase the computation for each time and enhance the efficiency of frequent item sets mining.In this paper, the author will focus on optimizing enumeration strategy and pruning strategy through the analysis and comparison of the advantages and disadvantages of the existing frequent item sets mining algorithms in two-dimensional and three-dimensional aspects. The author tries to improve the bitwise-based closed pattern mining algorithm BD-Miner in two-dimensional datasets and the bitwise-based closed pattern mining algorithm BD-Peeler in three-dimensional datasets. Both of the algorithms do not only inherit the advantages of existing algorithms, but also finish data mining tasks more efficiently.The BD-Miner and BD-Peeler is accomplished through the usage of VC++6.0 in this paper. A number of experiments have been carried out seriously, and the results have been compared with existing algorithms in the multiple datasets carefully. The experimental results show that, the efficiency of same data mining tasks in the same datasets and the same constraints can be improved 6-7 times in two-dimensional datasets with the use of BD-Miner algorithm; while 3 times in three-dimensional datasets with the BD-Peeler mining algorithm.