【作者】 李志国；

【导师】 任家东；

【作者基本信息】 燕山大学 ， 计算机应用技术， 2006， 硕士

【摘要】 数据仓库DW(Data Warehouse)的目的是要建立一种体系化的数据存储环境,将分析决策所需的大量数据从传统的操作环境中分离出来,使分散的、不一致的操作数据转换成集成的、统一的信息,企业内不同单位的成员都可以在此单一的环境之下,通过运用其中的数据与信息,发现全新的视野和新的问题、新的分析与想法,进而发展出制度化的决策系统,并获取更多经营效益。当前数据仓库研究的热点问题主要有:主题的设定;实化视图的选择;实化视图的维护;联机分析处理OLAP(On-Line Analytic processing);联机数据挖掘OLAM(On-Line Analytic and Mining);查询优化等等。首先,本文提出了一种新的主题搜索算法SSVC(Subject Searching Based on Characteristic Value)。该算法能够比较客观的搜索出最具有主题特征的表。该方法通过和需求分析得到的主题进行比较,判断需求分析和设计的客观性,帮助设计人员更有效的设计主题。另外,还可以对未知的关系数据库进行快速分析,找出最具有数据仓库主题特征的表,有效的辅助数据仓库设计人员确定主题事实表。其次,提出了一个新的选择实化视图的价值模型,并在该模型的基础上提出了实化视图选择算法PBPUS。该算法通过预处理得到候选视图集合,这样做的原因是缩小实化视图选择的空间。在维数很多的情况下,可以大大减少视图代价的计算量;然后根据新的价值模型计算出候选视图集合中视图的代价,利用改进的BPUS算法选出应该被实化的视图。新的价值模型通过考虑视图的查询频度和更新频度,再结合时间因素和空间因素,对原有的价值模型进行了改进。该算法与原有算法相比,降低了视图搜索的时间,减少了实化视图更新维护的代价,提高了实化视图的查询效率。再次,提出了基于时间戳的动态视图维护算法TS-DMV(Dynamic Maintenance Technique of Materialized View Based on Time Stamp)。该技术采用版本链控制技术,通过时间戳的控制进一步使视图更新和查询的同步进行,有效地解决了由于OLTP更新事务和OLAP事务同时访问数据所发生冲突的问题,在满足视图联机实时维护的同时,更好的提高了数据仓库的新鲜度和OLAP的查询效率。实验结果表明,本文提出的三个算法优于现有的同类算法,实现了预期的研究目标。更多还原

【Abstract】 Warehousing is an emerging technique for retrieval and integration of data from distributed, autonomous, possibly heterogeneous, information sources. Data warehouse is a repository of integrated information drawn form remote data sources, available for queries and analysis. By using the data and information from the data warehouse, we can obtain more benefit. Currently, it has many hotspots of data warehouse, such as building the subjects, selecting the materialized views, maintaining the materialized views, On-Line Analytical Processing, On-Line Analytical Processing and Data Mining and optimizing queries.Firstly, a new arithmetic for searching subject in data warehouse is proposed. It could objectively choose tables, which are fit for the subjects. By comparing with the subjects obtained by requirement analyzing, the new arithmetic is proved right. In addition, it could find out the tables fit for subjects by analyzing unknown database and assist designers to make fact tables.Secondly, a new cost value model is proposed. Based on this model, the arithmetic for selecting the views which should be materialized is brought forward. It could reduce effectively the amount of views by pre-processing. The new model considers query frequency and update frequency. Comparing with inhere arithmetic, the time complexity and maintenance cost is decreased, and the query efficiency of materialized views are improved.Thirdly, dynamic view maintenance technique based on timestamp is proposed, which adopts version-control technique. It preserves the synchronization of the querying and updating views by controlling the timestamp, which effectively solves the conflict of OLAP updating affair andOLAP accessing to database at the same time. Besides, it does not only satisfy the maintenance of views on-line real time, but also improves the freshness of data warehouse and querying efficiency of OLAP.Experimental results show that the algorithms proposed in this paper are more efficient than the current ones, and the anticipated results are realized.更多还原