Title: | 商业智能及其实验系统SmartBusiness |
Author: | 周艳
|
Issued Date: | 1999
|
Major: | 计算机软件
|
Degree Grantor: | 中国科学院中科院软件研究所
|
Place of Degree Grantor: | 中科院软件研究所
|
Degree Level: | 博士
|
Keyword: | 商业智能
; 数据仓库
; 在线分析
; 数据挖掘
; 语义层
|
Abstract: | 本文从商业智能的由来和发展说起,阐述了商业智能的概念、结构以及它的发展现状,它是从包括产品、客户、销售和其它业务信息中提取信息,并用先进的分析方法开发智能化的,基于事实依据的业务决策和业务战略的一门技术。然后详细介绍了实验系统SmartBusiness的设计思想、系统结构以及实现。SmartBusiness基于语义层技术,集查询、报表、在线分析(OLAP)与数据控掘(Data Mining)于一体,提供预订和发布功能,希望能为商业用户提供完善、丰富的决策支持工具。语义层技术提供数据库的业务术语表达,它就象一个透镜一样,通过它,用户能看到数据库中的数据,将商业用户与数据库技术术语和复杂的数据访问语言(如SQL)屏蔽开来,提供了统一的元数据定义方式。查询、报表基于语义层技术,把用户基于语义层之上的请求转换成标准的SQL语名,然后直接查询后台RDBMS获取结果。在线分析一般使用多维数据模型,本文介绍了三种典型的OLAP工具结构,并说明了他们各自的优缺点,SmartBusiness采用的是MQE这种方式,简洁而灵活,并注意克服这种方式带来的某些缺点。数据控掘针对不同的需求可以采用不同的数据挖掘模型和数据挖掘技术,SmartBusiness中暂时只融入了规则发现(Rules Discovery)和NN 预测(Nearest Neighbor Predict)这两种数据挖掘技术。规则发现有一系列的评判标准,如何设计好算法是一个值得研究的领域。NN预测有各种各样的算法,本文探讨了一些算法中存在的问题并说明了实验系统中采用的方式。接着,本文还就实验系统的实现、环境和管理方式作了一些简单的说明,最后的讨论指明了SmartBusiness中存在的问题以及需要改进和继续研究的几个方面。 |
English Abstract: | This thesis starts from the origin and development process of Business Intelligence and talks about the definition, structure and the status quo. Then we present the experimental system which is named "SmartBusiness". The system is based on the technology of semantic layer and it includes the functions on query and report, OLAP, data mining besides schedule and publish functions for users. The semantic layer presents the data in the RDBMS in a business way. It is a kind of metadata which is like lens so that people can look over the database without the knowledge on SQL. The query and report part translate the users' requests based on semantic layer to standard SQL statements. Thus they may query the RDBMS without any knowledge of SQL. The OLAP part introduces multidimenal model and three kinds of OLAP tools structure with advantages and disadvantages. SmartBusiness relies on the kind of MQE because it is smart and flexible. And we make efforts to lesson some of the disadvantages. Data mining is not a standard process. it has different kinds of processes and technologies. We should use different technologies according to different applications. SmartBusiness includes two mining technologies (Rules and Nearest Neighbors Predict) for the moment. There are some important measures by which rules should be presented to the end user. Better algorithms need furture research. Also we introduce the implement and management of SmartBusiness. Al last we bring up several questions on further improvement and research. |
Language: | 中文
|
Content Type: | 学位论文
|
URI: | http://ir.iscas.ac.cn/handle/311060/6508
|
Appears in Collections: | 中科院软件所
|
File Name/ File Size |
Content Type |
Version |
Access |
License |
|
LW002868.pdf(2147KB) | -- | -- | 限制开放 | -- | 联系获取全文 |
|
Recommended Citation: |
周艳. 商业智能及其实验系统SmartBusiness[D]. 中科院软件研究所. 中国科学院中科院软件研究所. 1999-01-01.
|
|
|