【作者】 陈洪顺；

【导师】 戚桂杰；

【作者基本信息】 山东大学 ， 管理科学与工程， 2011， 博士

【摘要】 随着计算机技术与网络技术的日益普及,数据库技术在各行各业得到广泛应用,同时也保存着大量数据信息,这些数据大多是面向应用的、低层次的,对于企业管理人员来说从中获取有帮助的知识是非常困难的,因为他们通常面对的是海量数据,难以从中提取到他们想要的知识,但这恰恰是他们最想要的,因此他们不得不处理大量数据来获取所需的知识,而这对管理人员来说,由于缺乏必需的技能和工具,往往难以取得这些知识。商业智能正是解决这一问题的有效的方法,它能够有效地处理大量数据信息,并从中获得有关知识。本论文以微软SQLServer的商业智能为基础,对其工作机制等方面进行了详尽而全面的分析研究,开发了一个商业智能应用平台；同时也分析研究数据仓库、OLAP、数据挖掘算法的使用及其在财务审计方面的应用。最后研究自定义数据挖掘算法的方法,以扩展商业智能的功能。本论文对于构造更强大的数据挖掘应用平台及其进一步的深入研究,具有一定的现实意义和参考价值。其创新点主要体现在以下几个方面：(1)理论框架上的创新：本文把商业智能理论与财务审计理论有机的融合在一起,形成一种新的、革命性的审计理论和手段。基于商业智能的计算机审计是在信息化环境下的一种崭新的审计方式。本文运用学科交叉和不同的思维方式去研究问题的不同层面和侧面,构建了基于商业智能的计算机审计的基本理论框架,提出了商业智能的一般模型,创建了基于商业智能的计算机审计系统的总体结构。(2)研究方法上的创新：提出并设计了基于商业智能架构的审计分析系统的框架,为审计技术研究提供了新的思路。分析了需要解决的问题,构建了性价比较高的、使用简便的、无需过多专业知识的商业智能审计分析系统,使审计人员从复杂的操作解脱出来,专注于审计理论与方法的研究。(3)研究理论的有效性：研究了孤立点挖掘算法在商业智能审计中的应用,以孤立点算法为工具进行审计中问题线索的发现。分析了孤立点挖掘与审计的关系,建立了孤立点挖掘模型,构建了孤立点挖掘审计线索发现模型,进行了数据实验,表明了方法的有效性。建立了收入成本配比审计模型,进行模型实验,结果表明了模型的有效性。(4)研究商业智能展示的科学性、直观性和可操作性：对一些重要的指标数据、审计分析结果、疑点数据以列表、图形、仪表盘等多种形式进行展现,并提供在机构、指标层次等多角度的灵活钻取和分析监测,使本文理论观点更具可靠性,指导具体实践活动更具可操作性和实用性。更多还原

【Abstract】 With the growing popularity of computer and network, database technologies are widely used in various industries, furthermore, a lot of data information is saved. These data are mostly application-oriented, low-level which is difficult for business executives to get knowledge because they usually face a mass of data and it is difficult for them to extract the knowledge they want, so they have to deal with large amounts of data. Business Intelligence is the effective way to solve this problem, since it can effectively handle large amounts of data information and obtain relevant knowledge.In this thesis, we base on Microsoft SQL Server business intelligence, with the comprehensive research of their working mechanism, we developed a business intelligence application platform, and meanwhile, we analyze date warehouse, OLAP, the use of data mining algorithms and its application in financial auditing. Finally, we discuss the way of custom data mining algorithms to expand the capabilities of business intelligence.This paper has some practical significance and reference value for the construction of more powerful data mining application platform and its further research. Its innovation is mainly embodied as following,(1) The innovation on the theory frame. This text compromises business intelligence theory and finance audit theory to form a kind of new and revoluting audit theory and method. In view of that the business intelligence audit method is a kind of newness in the information-based environment, this text applies academics cross and different modes of thinking to study the questions of different levels and sides and set up the basic theory frame based on business intelligence audit, and proposes the general pattern of business intelligence, and furthermore, establishes the total structure of the audit system.(2) The innovation of study method. This text proposes and designs the frame of audit analysis system and provides a new approach for audit techniques research. In this paper, we analyze the problems which needs solving and set up the business intelligence audit analysis system of high cost performance, simple-used and no need of more expertise, which frees the auditors from complicated operations and makes them focus on the research of audit theory and method.(3) The effectiveness of study theory. In this paper, we study the application of the algorithms of outlier detection mining in business intelligence audit and use the algorithms of outlier detection to find the problems of audit. We analyze the relationship between the outlier detection mining audits, and set up the pattern of the outlier detection mining, build the track discovery pattern of the outlier detection mining audit and have done some pattern experiments, which prove the effectiveness of method. We build audit pattern of cost and revenue matching with more experiments, which prove the effectiveness of pattern.(4) The study of the scientificity, visuality and operability of the business intelligence performance. We reveal some important index data, audit analysis result and doubt data in the forms of lists, pattern and instruments panel, etc. and provide analysis monitoring survey with the research of more angles study such as establishment and index levels, which makes the viewpoint of this paper more reliable and prove the operability and practicability of guidance in normal practices.更多还原