Association Rules for Auditing Systems

In this paper, we apply the association rules in data mining to an auditing system in order to mine the characteristics of audit data. The approach as a new mining technology can be used by an auditor to better interpret vast amounts of audit data. Association rules based algorithm is an outstanding methodology with which people can discover the hidden correlation relationships among dataset. It is applicable to mining of huge data which were difficult to start with. Because audit data usually contain a large number of rare data with different distribution characteristics, we hereby propose a multiple supports-based framework for digging data pattern from the rare data. We use all-confidence method to deal with crossing platform supports. In this paper we propose the MSAC_Apriori algorithm with generalized association rules, which helps establish the relationships during quantitative association analysis. Experimental results on practical datasets show that the proposed approach improves the performance by decreasing the number of frequent items without missing rare items.