【作者】 印莹；

【导师】 鲍玉斌；

【作者基本信息】 东北大学 ， 计算机应用技术， 2005， 硕士

【摘要】 OLAP技术是决策支持系统中的一种重要技术,用于管理人员决策分析。为了提高响应速度,需要对数据立方进行预计算。数据立方的计算在数据仓库中是非常必要但代价很大的操作。有效的数据立方(CUBE)计算成为研究的热点之一。 CUBE计算是OLAP即席查询分析的基础。CUBE操作内在的困难是计算代价和存储大小问题。一些文献提出了对CUBE计算的操作,包括并行性的应用、字符型映像到整型等,都是为了减少存储。但是随着新的应用,问题加剧了。比如,维数的增加和由此导致的CUBE大小爆炸性增长问题。其它文献已经提出了一些办法解决存储大小问题,而压缩的CUBE能同时减少计算时间和占用的空间。 Dwarf数据立方是一种高度的压缩结构,同时保持CUBE的语义,使OLAP查询易于实现。Dwarf数据方体将具有相同前缀和后缀的数据元素(Cell)压缩存储在一起。一般来说,数据方体在数据密集的地方前缀冗余比较多,数据稀疏的地方后缀冗余比较多。Dwarf数据方体存储时,消除了这两种类型的冗余,大大缩减了数据方体的存储空间。将一个完全实例化的数据方体缩减到一个非常紧凑的数据结构中。 本文研究了基于Dwarf的语义CUBE压缩技术。针对现有的Dwarf结构在实现过程中遇到的频繁I/O访问问题提出了改进算法—Q-Dwarf。另外,在实现基本Dwarf数据立方的基础上,进一步研究了以下几个问题。(1)为了更好的支持范围查询,研究并实现了聚簇算法。(2)实现了Dwarf数据立方的增量维护,研究并实现了基于三角形的增量更新算法。(3)为了更好的支持OLAP操作,研究并实现了基于文件的索引技术,实现其点查询和范围查询。大量实验表明,Q-Dwarf算法较原算法性能有明显提高,而且无论是在元组数较多的情况下还是在维数较多的情况下,我们算法的实现性能都较好。更多还原

【Abstract】 OLAP technology is very important in decision support system, which is used to aid knowledge worker’ decision and analysis. In order to reduce the query response time, we ought to compute the CUBE in advance. The computation of data CUBE is necessary but high cost in data warehouse. It is becoming one of the research focuses to compute data CUBE efficiency in data warehouse.The computation of CUBE is the basis of OLAP query and analysis. The inherent difficulty with the cube operation is the size, both for computing and storing it. Some researchers have provided some useful cube computation including the use of parallelism, and mapping string dimension types to integers for reducing the storage. The problem is exacerbated by the fact that new applications include an increasing number of dimensions and, thus, the explosion on the size of the cube is a real problem. All methods proposed in some literature try to deal with the space problem. The storage will be reduced dramatically by compressing data CUBE while the computation time can be saved efficiently.Dwarf is a highly compressed structure with OLAP semantic held. Dwarf solves the storage space problem, by identifying prefix and suffix structural redundancies in the structure of the cube and factors them out by coalescing their store. Generally, prefix redundancy is high on dense areas of cubes but suffix redundancy is significantly higher for sparse areas. Putting the two together fuses the exponential sizes of high dimensional full cubes into a dramatically condensed data structure.This thesis studies the issues of compressing semantic cube based on Dwarf. An improved algorithm—Q-Dwarf is proposed to solve the question of I/O frequently. Moreover, we go deeper into several issues after implementing the basic Dwarf CUBE. Firstly, the auxiliary cluster algorithm is implemented in更多还原