Research And Application Of Association Rules Technology Oriented Networking Audit Platform

After entering the 21 st century,our country's audit work shows effective supervision on the country's economic activities,improves the economic benefits of society,maintains a harmonious socialist society and many other benefits.In audit institutions,the computer technology has been applied widely and got more auditors' attention,because of the powerful capability of data query and analysis function.When aware of the importance of the computer technology,the national audit office issued the audit plan and the audit method system for each audit area.They are being adopted as auditing industry standards by audit institutions at each levels.It often cannot meet the needs of complex audit work because the audit method system is only simple SQL query operation written according to the laws and regulations.Having adopted advanced algorithm thought,data mining technology in recent years has obtained the amazing results after the application in the commercial,medical and other fields,thus has received the widespread attention.Because huge amounts of data should be processed,audit work calls for more data mining technology application in the field of audit,to improve the efficiency and quality of audit work,which has become an urgent request of the auditors.In this paper,the association rules in data mining technology is introduced into the auditing work,to provide a new means to analysis vast amounts of audit data for the auditors.Association rules algorithm is an unsupervised algorithm,and can discover hidden correlation rules between data items.It is very applicable to audit work where large amounts of data need to analyze and difficult to start with.Because the audit data contains a large number of rare data items,and has the tilt characteristics of the distribution.This paper proposed the multiple minimum supports framework and all-confidence metric to solve the problem.In this paper,the MSAC_Apriori algorithm of generalized association rules is proposed,to discover feature subset from quantitative data.In practice,it is hard to set minimum support of data item in multiple minimum supports framework.This paper proposed the method of deviation to set the minimum support.Experimental results on real world datasets show that the proposed approach improves performance well by decreasing the explosion of number of frequent itemsets without missing the frequent itemsets involving rare items.