Intelligent Audit System Based On Data Mining The Design And Realization

With the rapid development of the information technology and the network communication technology, the Computer-aided Audit System has been widely used. But during recent years, the information overload present a challenge for audit system to handle all the data. Adopting Data Mining to find out latent mode behind the data is the way to solve the problem.Data mining technology, including research background and related development, are summarized in the thesis at first. Then, based on the understanding of computer audit operation, a design model of the audit support system by using Data Mining technology is given, which including process of data collecting and data preparing. auditing data analyzing. The analytical method with union of fuzzy sets, neural networks and genetic algorithm on the data mining is discussed in this thesis. Concise rules and knowledge can be extracted using a fuzzy neutral network and genetic algorithm from database via following steps. Firstly, we set up of fuzzy neural networks, test sets and training sets, and meets the required accuracy rate of training. Secondly, right weights are optimized for genetic algorithm. Thirdly, redundant connections of the fuzzy neural networks are removed. Fourthly, the clustering of the activation values of the hidden units in the network is analyzed. Fifthly, audit rules are extracted. The analysis model is demonstrated effectively by applying it to some real data. This paper associates the practical condition of audit institution performing finance audit with the current state of domestic exertion of data warehouse, OLAP (On-Line Analysis Processing) analysis and data mining, introduces a solution that builds data warehouse on the loan business database system of bank and adopts OLAP analysis and data mining to audit unhealthy loan business. We will conduct bank non-performing loans audit and financial audit of the data analysis. Finally we compare the results of audit algorithm with the experience on the effectiveness. After verification, the audit model and method are feasible.