【作者】 崔家保；

【导师】 向阳；

【作者基本信息】 山东科技大学 ， 管理科学与工程， 2004， 硕士

【摘要】 随着税务部门信息化建设水平的不断提高，不同时间上线、不同开发商开发、不同业务部门使用的各种税务管理信息系统产生了大量的数据。然而这些数据却成了一些“信息孤岛”，缺乏有效的集成，它们很难为管理层的决策支持作出比查询更多的贡献。在这种情况下，本文讨论税务建设税务数据仓库和数据挖掘系统，以便能够好地为税务管理人员服务，提高其决策质量和效率。 本论文在熟悉税收业务和现有税务管理信息系统的基础上，探讨在税务系统建设税务数据仓库的方法和步骤，设计了税务数据仓库的数据模型和体系结构，建立了税务数据仓库与数据挖掘系统，并在数据仓库的基础上实现了数据的深层次分析。 总结本论文的研究内容，具有如下创新： 1、在数据仓库实现过程中，本文实现了基于时间戳的数据增量更新方法，该方法很好地解决了在数据仓库的海量数据中进行数据更新的问题。 2、以Analysis Services为OLAP引擎，本文借助ASP.NET、ADOMD、MDX、TeeChart等先进技术设计并实现了基于Web的OLAP数据展现和图形展现，可以为用户展示直观的报表和图形。 3、在对关联规则深入学习的基础上，提出了一种改进的Apriori算法，并在税务数据仓库中得到应用，该算法的创新之处在于：计算频度的方式、连接算法。 本论文中计算某个项目集出现频度的方法：不仅支持基于数据记录个数的频度计数方法，还支持用户自定义的频度计算方法，如Sum（），log（）等。 本论文中由k-1维频繁项目集生成k维候选集的连接算法：L_k-18 L_k-1={前k-2项都相同，并且第k-1项不同；同时两个作连接的项目集的第k-1项不同时出现在同一个代码表中，即这两个单项不具有互斥关系}。 4、以改进的Aprioi算法为基础，设计并实现了一个允许用户自定义的通用关联规则挖掘系统。更多还原

【Abstract】 With the rapid progress in the ’building of information, a tremendous amount of data has been generated in taxation bureau. However, these data are place distributed, meaning differentiated, it is hard to turn these data into useful information. They need to be integrated, transformed, cleaned, extracted and loaded to data warehouse which help to change the data into ’knowledge nuggets’. And organizations need the suitable knowledge to operate their business. Building a data warehouse in taxation bureau can help to serve managers more conveniently to make up their decisions.The thesis did some research on how to build taxation data warehouse on the basis of being familiar to the taxation business and it’s present information systems. Further, we design the data model and the architecture of taxation data warehouse. On the basis of data warehouse, we design the OLAP system and association rules mining system in order to make deeper data analysis.Some valuable works in the paper include:1. We worked out a new way to update the data warehouse based on inserting timestamp into the database. Many experiments have been done to achieve this new way.2. Analysis Services acts as the server of OLAP which provides capability and good performance and programming interface. Based on the techniques such as ASP.NET, ADOMD, MDX, and TeeChart, we devise a system to present data by tables and graphs based on Web.3. The paper put forward a association rule mining system which allows users to define their mining tasks freely. Meanwhile, some improvements have been done to make the system more efficient:(1) New aggregate functions have been devised,(2) A new prune way was used to make the algorithm more efficient according to the practical mining tasks, such as task relevant data.更多还原