Research And Application Of Audit Characteristics Intelligence Based On Association Rule Mining

With the rapid development of Intranet and database technologies, enterprises gradually construct Intranet environments and database application systems, day and day, each has stored large number of data. How to use those abundant data effectively to discover useful knowledge and rules that are unknown before, potential and useful is just the task of data mining. This function of data mining is also fit for the audit need. Audit study based on data mining is becoming a hotspot in the field of audit.1. The paper is an important part and one of key techniques of CNAP (Custom Network Audit Platform). Based on CNAP, an ARAC (Association Rule Audit Characteristic) is brought forward. ARAC focuses on data pretreatment with different structure before operation affair modes are created and intelligence is distilled based on association rule mining.2. The paper introduces XML to describe data with different structure and settle data integration problem. Data pretreatment is processed based on it. The model and arithmetics for data pretreatment using domain knowledge are designed in ARAC. Operation affair modes with altitudinal veracity are formed using operation affair modes recognizing arithmetic based on most frontad accessing path.3. The paper studies classical Apriori arithmetic. According to the problem of Apriori arithmetic, B -Gen( LK-1) arithmetic that generates candidate frequency itemsets based on binary and B - Calculate-count(C k) arithmetic that accounts support number are brought forward scientifically .The two methods have applied to CNAP system successfully and efficiently.4.In the end, based on the previous result, the running result of correlative experiment is contrasted and analysed, the feasibility and efficiency of the improved arithmetic and ARAC model are proved.Data mining of association rule technology is still a new application in audit field, and many technologies involved are not yet mature and being developed .The further research proplems are also discussed in the end of the paper.