Research Of Association Rules Mining Algorithms Based On Relational Database

With the rapid development of database technology, the data reserves all over the world have been enlarged in a quick format, behind which hide much potential information themselvs. However, because the technology carries no thorough analysis of data, the phenomenon "data rich but information poor" appears. Facing this challenge, the data mining technology declares its appearance in the certain situation. Association Rules Mining, which has widespread application, is an important matter in the data mining. In the beginning, association mining just focuses on business database. In recent years, the relational database obtained the widespread application, and studies of the effective technology, which carries on the association rules excavation in the relational database, supposed to be with futures so bright with promise.Nowadays, to excavate connection rule in relational database----For the first stage, relational database is transferred to be business database by the way of changing Quantification attribute in the former on into Boolean attribute, for the second step, it is Boolean excavation algorithm that used to mine. This method is an efficient one because Boolean is connected the excavation algorithm. However, for most methods available, after transformation, Apriori algorithm or its variation if applied directly. In that case, the speciality of association rules is ignored. As a result, the efficiency of mining is reduced.Considering the above, this paper makes a careful analysis of speciality of association rules in relational database, and presents a new algorithm of association rules excavation based on array, on the view of Apriori algorithm. In this new algorithm, Multi-dimensional association restraint, which can not be operated in former Apriori algorithm, is applied in "Pruning". Furthermore, the number of candidate item of collection is decreased, and the only one time database scan cuts I/O expenses. This new algorithm has been applied to the data mining of The Criterion of Healthy Physique in Anhui Science and Technology University, and the experiment results show that this algorithm is quick, effective, easy to develop.Usually, data mining is used to process massive data. This paper is a thorough research both on how to apply this new algorithm to the data mining and on strategies based on the category.The evaluation of association rules is a key stage to a successful association mining. This paper is a thorough research both on how to apply this new algorithm to the data mining and on strategies based on the category.The evaluation of association rules is a key stage to a successful association mining. Although in most Association Rule Mining Algorithm, the application of smallest support and smallest confidence avoids certain dull rules, most rules, uninteresting, even false, can not be wiped away. At the end of this paper, the evaluation of association rules is supposed to a thorough discussion.