The Research On Mining Algorithm Of Association Rules Based On Attributes Covering

Data mining is a new subject involving multidisciplinary fields, in which data mining of the association rules is a very important research branch. The objects of such a research is always a database containing massive raw data with the large number of items. It is an important research problem to design an efficient algorithm to improve the efficiency of data mining of the association rules.The major research work of this dissertation is as follows:The concept of traditional data mining of the association rules has been expanded. And the concepts such as frequent item set, attributes covering based on the different transactions and so on has been introduced.An algorithm of association rules based on attributes covering has been put forward on the basis of the research of the classical algorithms of the association rules. This algorithm scans the original database only once to get the compressed link-list-database, and the subsequent process of the data mining is carried out in it, so it has the higher efficiency. And the algorithm does not produce candidate item sets, thus avoiding the combinatorial explosion of the candidate item sets.Finally this dissertation puts forward the stitching tree model based on the ArBac algorithm in order to make the algorithm applicable to the data mining in the large transactions database.