摘要: |
由于高维属性和海量数据所带来的影响,数据管理需要相当高的计算负载,传统的
集中索引技术已经变得不切实际。为满足数据的快速增长、海量和高维特性的要求,实现了
一个高层次的分布式树形索引结构框架MRC-Tree。基于MRC-
Tree框架基础上,提出了两种MKd-Tree索引结构构建方法,即OMKd-Tree和MMKd-Tree。
理论分析和实验结果表明,基于MRC-Tree框架的MKd-Tree索引结构构建方法具有良好的可扩
展性和较高的检索效率。 |
关键词: 高维数据库;图数据;索引结构 分布式树形索引结构框架 Map-Reduce框架 MKd-Tree |
DOI:10.3969/j.issn.1001-893x.2013.07.017 |
|
基金项目: |
|
Large scale graph data index based on MKd-Tree in cloud environment |
LEI Ting |
() |
Abstract: |
Managing the high-dimensional, large-scale data needs extremely high co
mputational load. Traditional centralized indexing techniques apparently become
impractical. To address the demanding needs caused by this rapidly growing, larg
e-scale, and high-dimensional information ecology,a high-level di
stributed framework for searches and computations on tree indexing structures ba
sed on Map-Reduce in the Hadoop environment, MRC-Tree (Computation based on Ma
p-Reduce on tree structures) is achieved. And then,two MKd-Tree(Kd-Tree base
d on Map-Reduce) index structures based on MRC-Tree framework, OMKd-Tree (Bu
ild one distributed Kd-Tree based on Map-Reduce) and MMKd-tree (Build multipl
e Kd-Trees by splitting data equally based on Map-Reduce) are proposed. Finall
y, the theoretical analysis and experiment results illustrate that the methods are highly eff
ective and extensible to the similarity search in high-dimensional data environ
ment. |
Key words: high-demensional database graph data index structure distributed tree index structure framework Map-Reduce framework MKd-Tree |