Font Size: a A A

The Research And Application Of Mining Theory On Association Rules

Posted on:2007-01-07Degree:MasterType:Thesis
Country:ChinaCandidate:J XiongFull Text:PDF
GTID:2178360185461051Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the developing of the information process, data resource increases fast and become rich, then how to find the useful imformations from the data resource becomes very important. The data mining technique appeared and now becomes one of the advanced study directions in the field of database and intelligent decision support system.Mining association rules is one of the important problems in data mining. However, many algorithms have not considered this aspect enough that mining data between people characters and their actions is also an important aspect for multidimensional association rules. In this paper, a multidimensional association rules mining algorithm on the basis of dimension restricted and hashing is presented. Its whole name is Count of candidates and Prune transactions based on Hashing and it is called CPH for short. And reviewing attributes' correlation, an association rules mining algorithm on opposite attributes is presented. Its whole name is Association rules of opposite Attribute Aggregates and it is called AOAA for short. The two algorithms are applied in the integrated network management system for the opening large-scale laboratory.The paper consists seven chapters. The research of association rules mining algorithms, the improvement and application of the multidimensional association rules mining algorithm are emphases. The primary studys and outcomes are shown below. The paper introduces the data mining technique, compares many data mining algorithms and studys the classical and improved algorithms on association rules. According to the analysis of association rules mining algorithms, the CPH and AOAA algorithms are presented. The CPH algorithm adopts dimension search, hashing and compressing transaction techniques, reduces the times of scanning the database and enhances the efficiency to discover multidimensional frequent itemset. The AOAA algorithm considers attributes' correlation to discuss the opposite attributes' association rules and it enhances the rules' confidence. The two algorithms are applied in the integrated network management system for the opening large-scale laboratory. According to the result of analyzing and validating the performance of the algorithms, it shows the algorithms have the better performance.
Keywords/Search Tags:association rules, frequent itemset, dimension restricted, Hashing, CPH
PDF Full Text Request
Related items