【作者】 吴璇；

【导师】 李敏强；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 随着信息技术的迅猛发展,全球性信息化迅速到来,经济领域高科技成果的层出不穷,给审计信息化进程带来一定的压力和动力。本项研究以人工智能理论与技术、审计理论技术为研究共同基础,采用了理论分析和实证检验相结合的研究方法,重点研究了基于人工智能方法的金审工程问题,较为完备的研究了智能化审计总体系统、智能化内部控制评价系统、内部控制评价中的定性仿真以及神经网络在审计智能中的应用。主要研究内容和创新成果如下:(1)首次把智能理论技术与审计理论技术进行了融合。计算机审计是在信息化环境下的一门新的审计学科,是一种崭新的审计方式。构建了计算机审计的基本理论框架,提出了计算机审计的一般模型,主要包括计算机环境下的审计对象、审计作业模式、审计基本方式、审计技术方法等。提出了智能化审计系统的总体结构,包括现场审计实施系统、联网审计实施系统、审计办公系统和信息资源库四个部分。分析了数据采集与转换,包括数据采集与分析的基本特征,数据采集智能化,并构建了审计数据采集与转换系统。(2)提出并设计了内部控制智能评价系统。从知识角度总结了国外、国内内部控制制度和体系。分析内部控制需要解决的问题,设计了内部控制智能评价系统,包括系统的整体结构、系统的知识库、数据库、推理机制等。将内部控制与定性仿真方法相结合,研究了QSIM算法,设计了基于定性推理的内部控制评价系统结构,以及系统处理流程和系统功能结构等概念模型。把内部控制的业务循环进行了简化与抽象,并对业务循环的各系统模块进行了分解,建立定性的成因推理模型。(3)研究了人工神经网络在审计智能中的应用,以神经网络为工具进行审计中问题线索的发现。分析了数据挖掘、神经网络与审计的关系,借鉴了国外神经网络在审计方面的应用,建立了神经网络方法在审计分析性复核过程应用模式。选定BP神经网络为工具,构建了人工神经网络与财务危机发现模型,进行了数据实验,表明了方法的有效性。建立了基于自组织特征映射的纳税情况审计模型,税种选择为增值税,进行模型实验,结果表明了模型的有效性。更多还原

【Abstract】 With the rapid development of information technology, the global informationalization is imminent. High-tech achievements in the economic field are emerging one after another. It brings pressure and driving force to audit Information system. On the basis of artificial intelligence and audit, the thesis employs empirical study and academic analysis to study the Golden-Audit program, and designs the General System of Intelligence Audit, Intelligent Evaluating System of Internal Control, Qualitative Simulation in internal control. The major research work and achievements are as follows.The computing auditing and artificial intelligence methods are combined to deal with audit problems. Computing auditing is a new subject and audit technology in the information age. The framework for computing auditing is initially established, including auditee, audit operation model, audit basic fashion, audit new technology and so on in computing environment. It gives priorities to the transformation from computing auditing to intelligent audit system. The general structure of intelligent audit system is built, including conduct system on field work, on net, audit office system and information resource data. The investigation is made on the gathering and transformation of data, including intelligent gathering and transformation of data, and the system of the gathering and transformation of data is designed.The thesis proposes an artificial intelligent evaluation system for internal control. Firstly, the research gives a review of foreign and domestic internal control system. Secondly, we design a artificial intelligent evaluation system for internal control to analyze audit internal control problems. The analyses are focused on the design of system integral structure, knowledge base, database and reasoning mechanism which integrates internal control and qualitative simulation. By using the QSIM,the structure of intelligent evaluating system of internal control is designed based on the qualitative simulation. Meanwhile, the process and the functionalities of system are built with concept models. The transaction cycle of internal control is simplified and abstracted. The system module is decomposed and the qualitative reasoning modeled.The artificial neural network is applied to support auditing as an analytic tool to find the clue of problems. Firstly, we analyze the relationship between data mining, neural net and auditing. By referring to the applications of neural network, we find that the neural network is suitable for the analytical review process of audit. Secondly, we select BP as a tool to construct models of artificial neural network for identifying financial crises based on empirical data. We conduct an empirical study on the financial risk model of BP neural network and the rate paying audit model of SOM. The value-added tax is selected for experiments, which verifies the proposed model and its validity.更多还原