eXist数据库结构连接算法的改进和实现  

IMPROVING AND IMPLEMENTING STRUCTURAL JOIN ALGORITHM FOR EXIST DATABASE

作　　者：任涵[1] 潘荫荣[1] 胡幼华[1] 

机构地区：[1]华东师范大学计算机科学技术系,上海200062

出　　处：《计算机应用与软件》2010年第1期177-179,共3页Computer Applications and Software

摘　　要：为了进一步提高eXist数据库的查询效率,针对数据库现有的数据存储特点和查询策略,提出一种快速定位的结构连接算法FL-Stack。算法采用栈缓存已遍历过的但仍可能与尚未遍历的后代结点匹配的祖先结点,并对能预先判断不可能满足结构连接匹配的祖先或后代结点,提出相应方法实现快速定位,以批量跳过这类结点。与现有算法必须逐个扫描祖先后代结点序列相比,这种快速定位的结构连接算法避免了逐个扫描带来的多次重复无意义的比较,可大大提高结构连接效率。In order to improve query efficiency of eXist database,this paper provides a structural join algorithm featuring in fast locating（FL-Stack） based on current data storage and query strategies of the eXist database.The algorithm uses stack cache for traversed ancestor nodes of them there are still the possibility to be matched by descendant nodes without traversal,and is able to manage a corresponding approach for fast locating and bypassing in batch those ancestor or descendant nodes that they can be pre-judged of impossible to meet structural join matching.Current algorithm in eXist database has to scan the ancestor and descendant nodes＇series one by one,comparing with this,the structural join algorithm with fast locating is free from multiple repeated meaningless comparisons incurred from one-by-one scanning,and improves the structural join efficiency significantly.

关 键 词：XML eXist 结构连接 快速定位 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] P228.4[自动化与计算机技术—计算机科学与技术]