The Research Of Multi-dimensional And Multi-level Association Rule Algorithm

With the growing size of the database and prosperity of data mining technology, association rules technology also has been an explosion of development and are moving in the direction of more extensive and in-depth. Association rule mining algorithm for mining association rules is the main content of the study. The key to improve the efficiency of association rules is to improve the efficiency of association rules algorithm. Apriori algorithm is one of the most influential single-Jorgen Weibull-based association rules mining frequent itemsets algorithm. Apriori algorithm has two major bottlenecks exist: first is the number of candidate itemsets, and second is the number of scanning the transaction database. At the same time Apriori algorithm is a single-Jorgen Weibull type. Study of the classical association rules have been expanded from a one-dimensional single-level to multi-dimensional and multi-level mining. The concept of using an abstract level, you may find new, more abstract rules. In practice, it should be from different angles at different levels on the excavation, under these conditions strong association rules generated by the people to be more useful. Because many data are stored as the form of multi-dimensional data in a relational database. Therefore, this paper is to extend single-Jorgen Weibull-type algorithm Apriori algorithm for association rules to multi-dimensional multi-relational data mining up.On the basis of intensive research on association rules of data mining techniques, this paper has to do the following work:(1) Analysis of the classical algorithm Apriori association rules algorithm, including the method of thought, the algorithm main steps and algorithm pseudo-code, and analyzes its problems, listing some ways to increase the effectiveness of the method of Apriori.(2) In the full absorption of the classic Apriori algorithm has proposed the improved algorithm, the improved algorithm is suitable for multi-dimensional relational data mining. Mainly describes the improved algorithm, the idea of the algorithm, pseudo-code and algorithm correctness of the theoretical analysis.(3) Compared test has been used between improved performance of the algorithm and the Apriori algorithm. The result shows the advantages of the improved algorithm in multi-dimensional .Finally, this paper summarizes and discusses the further work.