RAKDB-Tree——一种基于近似区域的多维数据索引结构  

The RAKDB-Tree An Efficient Approximation-Based High-Dimensional Index Structure

作　　者：黄维辉 熊翱[1] 

机构地区：[1]北京邮电大学网络技术研究院,北京100876

出　　处：《软件》2013年第11期77-79,共3页Software

摘　　要：多维数据的处理已经成为影响很多领域发展的关键因素,特别是多维数据的相似性查询已经被用在很多领域中。当数据维度很大的时候,大多数索引结构处理的性能下降,这现象被称为"维度灾难"。针对多维度灾难,RAKDB-Tree是本文提出的一种高效处理多维数据的索引结构。该索引结构首先把数据空间划分为子空间,然后使用改进的KDB-Tree对子空间建立索引。RAKDB-Tree的查询、插入、删除等算法使得,索引结构一直保持较优状态。实验结果表明,RAKDB-Tree能够很好解决因为数据维度增加而带来的各种问题。For many application areas, the efficiency of multidimensional data processing is a key factor affecting their development. In particular, similarity search is used in many fields, such as data mining, big data analysis, digital multimedia etc. However, lots of index structures cannot avoid the ＂dimensionality curse＂, when number of dimensions is very large. RAKDB-Tree uses partitioning method to divide data regions and create approximation files. Then RAKDB-Tree indexes the approximations with the improved method of KDB-Tree. RAKDB-Tree is an automatically adjust and optimize tree index structure. Experimental results show that the RAKDB-Tree has a promising improvement in performance.

关 键 词：相似性查询 多维数据 索引结构 RAKDB-Tree 

分 类 号：TP311.134[自动化与计算机技术—计算机软件与理论]