基于区位码和距离的Chord网络高维数据范围检索  

Bit-code and distance-based range query for high-dimensional data in Chord network

作　　者：孟宪福[1] 孟泓汐[2] 张振强[1] 

机构地区：[1]大连理工大学计算机科学与技术学院,辽宁大连116024 [2]大连海事大学信息科学技术学院,辽宁大连116026

出　　处：《大连理工大学学报》2011年第3期433-439,共7页Journal of Dalian University of Technology

摘　　要：结构化对等网络对数据的范围检索等相似查询缺乏有效的支持.由于利用iDistance索引进行范围查询时会引入很多"误中点",提出了一种利用区位码和距离来建立高维数据索引的方法.该方法首先利用Code-Distance技术来建立高维数据的一维索引,并采用位置保持哈希函数给予每个索引值一个唯一标识,该标识将被保存在Chord环节点上,从而生成BM-Chord系统.同时,详细叙述了数据过滤技术和范围查询策略.模拟实验结果表明,BM-Chord系统在减小中间结果集大小和提高查全率等方面是有效的.A structured P2P network is not suitable for similarity search,such as range query.Since a lot of false hits may be obtained by using iDistance indexing technique,a high-dimensional data indexing algorithm based on bit-code and distance is proposed.The Code-Distance technique and locality-preserving hashing function are utilized to respectively construct one-dimensional index from data with high-dimension and assign a unique identifier for each index,and those identifiers are kept into nodes to establish BM-Chord network.The data filtering technique and range query strategy are described in detail as well.The simulation results testify the effectiveness of BM-Chord network in terms of the intermediate data set and recall ratio.

关 键 词：P2P 范围检索 iDistance 位置保持哈希函数 区位码 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]