节点文献

基于孤立点的数据挖掘研究及其在计算机审计系统中的应用

Research of Data Mining Based on Outliers and Appliction in It Audit System

【作者】 廖轶

【导师】 黄厚宽; 杨蕴毅;

【作者基本信息】 北京交通大学 , 计算机软件与理论, 2007, 硕士

【摘要】 自上世纪90年代以来,计算机、网络、通讯技术得到迅速发展,应用领域也从军事逐渐转向民用,以其广泛的影响和强大的生命力向经济社会、社会生活的各个方面扩展渗透。网络化对于审计工作带来的挑战也源于经济管理领域计算机,通讯技术的普遍应用。目前,国家机关、重点企事业单位的经济活动和会计信息电子化己经普及。特别是随着互联网的发展,电子商务、网上交易的产生,计算机审计已逐步走向网络,财务管理已突破时空间局限,实现了对异地机构财政财务收支活动的及时审计核查。审计对象和审计环境迅速发展变化,要求审计机关的作业方式必须与时俱进,以不同的审计应对不同的审计对象。计算机审计是审计领域的新生事物,代表了审计理论与实践的发展方向,具有传统的手工审计不可比拟的优势。建立计算机联网审计系统,可以提高审计工作的效率,进一步防范审计风险,提高审计工作的质量和管理水平。首先本文作者给出了“计算机审计”概念的定义,并通过对国内外计算机审计的现状进行分析,总结了当前计算机审计的研究热点。第二章介绍了孤立点发现的研究过程及当前研究动态,介绍了基于聚类、基于距离、基于统计、基于密度、基于偏离以及高维数据孤立点发现中的主要算法,具体分析了各个算法的主要内容,在此基础上总结比较了各个算法的优劣及其适用范围。第三章,在基于单元的孤立点发现算法的基础上,提出了一种减少边缘影响的孤立点分析算法。针对算法中边界处孤立点的误判问题,给出了数据空间的单元格划分及数据对象分配方法,定义了数据集边界阈值动态调整函数,提出了基于单元的孤立点挖掘算法的改进算法,在不增加原有算法时间复杂度的前提下,极大地减少了边界处孤立点的误判。第四章,本文通过介绍Benford法则,对Benford法则应用进行了重点论述。第五章,详细阐述了计算机联网审计系统,分析了三种不同的组网模式,并结合不同的交换处理技术,对其结果给出了详细的分析比较,还介绍了审计数据转换技术。第六章,运用Benford法则,进一步对数据进行过滤,得出了可疑的数据,并从这些孤立点中结合改进的基于单元的改进算法进行判断。从而提高了对孤立点判断的准确率。最后,对本文的工作进行了总结和对研究前景的展望。

【Abstract】 Since 90’s in last century, the calculator, network, the communication technique gets the quick development, applying the realm to also join the army the matter to change direction gradually public, with its extensive influence and strong vitalitiesare to the economic society, each one of the social activities expands to permeate. The networking also comes from the technical and widespread application of the economy management realm calculator, the communication to the challenge that the audit work bring. Currently, the economic activities the accountancy’s the information of the national organization,the point business unit are electronic to have already made widely available.Especially along with connect the development of the net with each other, creation settle accounts on the electronic commerce, IT audit have already headed for the network gradually, the finance management has already broken the time space to limit, carrying out to the instantly audit organization public finance receipt and expenditure activity.Computer auditing has many unmatched predominance compared with the traditional auditing, but the study of computer auditing is right at an incipient stage in our country. The Audit the object and audit the quick development of environment variety, the operation method that requests to audit the organization must with hour to enter, replying the different audit object with the different audit. Establishing IT networking audit, and the usage of the system can raise the efficiency of audit the work, further guarding against the audit risk, the exaltation audit the quantity of the work and manage the level.First , the author gives the definition of a concept“IT audit”, concludes some research hotspots on the analysis of national and international IT audit status quo.Computer auditing is the new comer in the field of auditing; it represents the trend of auditing theory as well as practice.Second, the research process and the current situation of outlier detection are reviewed. The algorithms of outlier detection based clustering, distance, density, stat.,deviation and high dimension are introduced. The content of these algorithms is analyzed. The disadvantages and advantages of these algorithms are compared.In the third chapter, based on the algorithm of based-cell outlier detection, an outlier-analysis algorithm to reduce the boundary influence is presented. The data

【关键词】 孤立点数据挖掘边界单元格Benford法则
【Key words】 OutlierData miningBoundary cellBenford law
  • 【分类号】TP311.13
  • 【被引频次】15
  • 【下载频次】788
节点文献中: 

本文链接的文献网络图示:

本文的引文网络