基于焦点和角度的多维索引方法

多维索引方法的算法非常复杂且难于实现,有时算法的复杂程度和其性能的提高是不相匹配的.为此,作者提出了一种基于焦点和角度的多维索引结构.基本思想是在对象空间选出焦点集,通过计算得到中心焦点、基本向量集和FAC＿坐标.在检索时,通过估计结果集内数据点与基本向量的夹角范围来实现对数据点的过滤.这种索引方法的最大优点是索引文件较小,所需的存储空间小.因而,这种方法能够更好的适应于维数和数据集的增长.此索引结构与Omni＿顺序扫描算法的过滤效率通过实验进行了对比,实验数据验证了该索引方法的有效性.

An Efficient Indexing Method for Multi-Dimensional Data Based on Foci and Angles

Many indexing approaches for multi-dimensional data have evolved into very complex algorithms which are hard to implement. Motivated by this situation, the author propose a simple yet efficient indexing method for multi-dimensional data which is based on foci and angles. The basic idea is to select a set of objects as foci. And the central focus , the basic vectors and FAC-coordinates can be got by computing. We can filter data objects by estimating the range of angles which are belonged in results set when we retrieve. The strongest point of this method is the small size of indexing file because the angles which are independent of dimensions need small storage space. So this method scales well for growing dimensions and database size as well as easy to implement. The results of experiments show us the efficiency of this method.