基于优化过滤策略的XML数据查询处理  

An Optimized Filter Strategy for XML Query Processing

作　　者：陈海坤[1] 李建中[1] 骆吉洲[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术系,哈尔滨150001

出　　处：《计算机科学》2007年第4期111-113,共3页Computer Science

基　　金：国家自然基金重点项目;编号60533110;黑龙江省自然科学基金重点项目;编号zjg03-05;国家自然科学基金项目;编号60473075;国家教育部新世纪创新人才计划;编号NCEF-05-0333;黑龙江省自然基金;编号F0208;哈尔滨市科技攻关项目;编号2004AA1CG13213

摘　　要：如何高效地处理XML查询,是目前研究的热点。由于当前方法存在过多扫描无用节点引起效率下降的问题,本文设计了一种XML数据的二级索引结构,基于该结构给出路径查询处理算法。首先,本文对XML模式中每个节点按路径类型进行分类编码,然后把每个节点按该编码进行聚类存储。在查询时,就可以先根据模式信息和查询信息得到目标节点的编码,然后只需将二级索引中这些编码对应的部分载入内存,进行过滤操作。这样就不必扫描整个索引,提高CPU和IO效率。本文还对二级索引结构进行扩展,使本文的过滤索引能方便应用在有分支结构的查询上。实验结果表明,本文的XML数据过滤算法效率优于基于Bitvector的过滤算法,并且索引结构所需要的存储空间也小于Bitvector索引。How to retrieve information user interested in has become a hotspot in research. In this paper, we propose a filter strategy to reduce the number of candidate nodes using double level index. In the filter algorithm, all nodes are coded according to their path type, and clustered based on their codes. In course of processing, we first calculate the target codes of the query according to the structure of XML data. Then we load parts of the index which target codes point to into memory, and use the filter expression to filter those definitely useless nodes without scanning entire in- dex. Furthermore, we also extend the index structure to process twig pattern query for XML. The analysis and the experimental results show that our filter algorithm has a better performance than filter algorithm using Bit vector, and also needs less storage.

关 键 词：XML 数据过滤 路径表达式 模式图 位向量 二级索引 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]