| Association rules have been recognized as a very important topic of data mining research. Since R.Agrawal derived the concept of association rules and proposed association rules mining algorithms firstly, a lot of researchers have broadly researched on association rules mining algorithms because it has commercial value and theoretic value. The difficulty of mining association rules is that the mining object is very large database. Because R.Agrawal's mining algorithm needs to scan repeatly and tediously the database, it is no actual value to the real database. The algorithm---- FP-Growth is about an order of magnitude faster than the Apriori , but the algorithm takes more memory,so there is difficulty to realize the algorithm in real applications involving very large database. At present there are a large number of bibliographies on the research of association rules inland and abroad, but most of them are focus on the improved algorithms based on the two algorithms,and the real and available algorithm hasn't appeared.The main content of this dissertation is about the algorithms of mining association rules and realization of system. In order to improve the algorithm's efficiency, I proposed a algorithm (A-Mine) basis of array and two kinds of incremental updating algorithms by studying the inner relationships of data in database and by using special data structure for the research of association rules mining algorithms. The results of the experiments show that the algorithms have been improved greatly in realization and efficiency by considering data structure and inner relationships.In this dissertation, I studied the mining association rules through the following three aspects and achived some consequence.1. Based on the research of correlative bibliographies about two kind of classical association rules mining algorithms (Apriori and FP-Growth), I proposed a association rules mining algorithm (A-Mine) on the basis of the array by using a simple data structure----array. |
PDF Full Text Request