面向动态审计轨迹的数据探索式预警系统研究

项目名称： 面向动态审计轨迹的数据探索式预警系统研究

项目编号： No.71271117

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 管理科学

项目作者： 陈耿

作者单位： 南京审计学院

项目金额： 54.5万元

中文摘要： 审计在本质上是国家经济社会运行的免疫系统，发挥着预防、揭示和抵御各种风险的功能。在分布式经济行为向网络空间映射的过程中，反映审计行为的数据呈现出多结构、多源化和超海量等大数据特征，导致现有的审计信息化技术与方法难以充分发挥审计的免疫系统功能。本课题从以下几个方面展开研究：跨域环境下审计数据的可信共享技术，包括协议交互可控模型及信任控制机制、协议交互可审模型及抵赖控制机制、可控可审联动模型与智能协同方法等；面向大数据的数据探索方法研究，包括面向复杂关联的深度分析、多结构处理、超海量数据管理等技术，以解决多结构挖掘、高阶有趣关联发现，以及数据挖掘的时效性问题；基于数据探索的审计预警模型构建方法，包括审计行为的评估模型构建、行为动态传播预测机制等；构建面向审计轨迹的审计预警原型系统，作为电子审计系统中发挥免疫系统功能的核心模块等。本课题探讨大数据环境下的审计行为，为审计信息化理论提供了新思路。

中文关键词： 大数据；审计预警；数据分析；可信协议；深度分析

英文摘要： The audit is essentially an immune system of national economics and society with the functions of preventing from, revealing of and resisting with various risks. During the course of the distributed economic behavior maps into cyberspace, the data reflecting the audit beheavors show as the big data with the characteristic of multi-structure, multi-source and super-massive, which results in the existing audit methods are difficult to give full play to the functions of the immune system of audit. With these issues in mind, we study the subject from the following aspects: Firstly, we study the technology of trusted sharing of audit data in the cross-domain environment, including the protocol controllablity model with the trust control mechanism, and the protocol auditability model with the non-repudiation control mechanism, and conjunction model of controllablity and auditability with the intelligent collaborativation mechanism. Secondly, in order to solve multi-structure mining and discover interesting association with high-end,and improve the timeliness of data mining, we study the data exploration methods for big data, including the deeply analysis technology oriented to complex association, multi-structure processing technology, and super-massive data management technology. Thirdly, we study the building method

英文关键词： big data；early-warning of audit；data analysis；trusted prococol；deep analysis