Research And Application On Association Rules Technology In The Audit Field

Association rules mining is an important research branch of data mining, which is usedto find the hidden association between the data item in the database. In order to extend theknowledge that can be expressed by traditional association rule, the thesis propose a methodof generalized association rules mining based on quantitative attribute, by combining researchemphasis of quantitative association rules and negative association rules. In order to find theassociation of in social security audit data items, the proposed method is applied in socialsecurity audit datasets, and then significance of association rule in the area of security audit.Firstly, the thesis introduces the research status of theory and application of associationrules mining, and then analysis the classic method of association rules mining. Inspired by theidea of study on quantitative association rules and negative association rules, the paperpropose the method of handle the dataset based on quantitative attribute efficiently, and thenpropose Apriori_PN algorithm which can mining positive and negative frequent item setssimultaneously. Extracting interesting rule from results of Apriori_PN algorithm mining basedon the parameter of interest measuring the rules Useful degree, based on the idea, positive andnegative association rules producing algorithm（GAR_I）is proposed. Apriori_PN algorithmand GAR_I algorithm are applied in the social security audit data. Firstly, the paper proposesthe discrimination algorithm DISC_S, which is suited for characteristic of social securityaudit data and preprocesses data by the discrimination algorithm. Secondly, association rulesare mined by Apriori_PN algorithm and GAR_I algorithm which is used to extend the auditknowledge base and propose evidence of analysis of audit.