【作者】 李军；

【导师】 刘先锋；

【作者基本信息】 湖南师范大学 ， 计算机软件与理论， 2009， 硕士

【摘要】 XML(eXtensible Markup Language)己成为因特网上的数据存储、交换和表示的标准。随着XML应用的普及,越来越多的数据以XML的形式存储和交换,对XML文档中的数据进行查询的要求也就越来越高。为充分利用XML的特点开展查询,本文详细研究了XML查询改进技术。MarcXchange文档是一种应用在数字化图书馆领域里的XML文档类型,可以作为通用的数据交换格式在Web环境下共享。为更好地促进数字图书馆建设,本文还详细研究了MarcXchange文档的索引和查询技术。本文的主要贡献有:1.提出了一种改进的基于LI-MOON编码的XML索引,当我们对XML编码后的索引文档进行B+树结构化处理后,能大大减少索引文档的大小,因而能有效提高查询速度。2.针对Stack-Tree-Desc算法没有利用索引改善连接效率,浏览了一些不必要的节点浪费了资源,影响了查询速度。本文提出了一种新的连接算法,通过B+树查找定位,来跳过祖先和后裔列表不必要的元素以提高查询速度。3.本文提出了一种基于MarcXchange文档的结构索引方法—Marc_index,该索引方法根据MarcXchange文档的特点采用LI-MOON编码方式,有效地实现了对文档树中任意两个结点之间祖先/后裔关系和双亲/孩子关系的检测。该索引另外的创新之处还在于有效结合B+树和倒排索引这两种方法,从而综合两者的优点,克服双方的缺陷以提高查询效率。最后,本文还提出了一个基于Marc_index索引的查询处理算法。更多还原

【Abstract】 XML(eXtensible Markup Language)has already been the standard of data storage,exchange and representation on the Internet.With the spread of XML application,more and more data is stored and exchanged in the form of XML.As a result,the demand for query data contained in the XML documents is becoming more and more critical.To fully apply the characters of XML to query,we present our contributions to the improvement techniques of XML query.MarcXchange document is a kind of XML document type for application in the digital library,which can be used as a common data exchange standard to share in the web environment.In order to promote the digital library construction,the paper studies in detail index and query technology of MarcXchange.1.This paper proposes an improved XML index based on LI-MOON code,when we encode indexed documents by B+ tree structure,It greatly reduces the size of indexed documents so that effectively improve query speed.2.For Stack-Tree-Desc algorithm improving the connection efficiency without index and browsing a number of unnecessary nodes to waste resources and affect the query speed.A new connection algorithm is proposed for improving query speed that it searches position by B+ tree index to skip unnecessary element in the list of ancestors and descendants3.Marc_index based on MarcXchange documents is proposed that is using LI-MOON code in accordance with the characteristics of the MarcXchange documents,achieves effectively detection of two nodes ancestor/descendant relations and parent/child relations in the document tree.The other innovation of the index is that it integrate effectively advantages both B+ tree and inverted index to overcome the shortcomings in order to improve query efficiency.Finally,the paper also proposed a query processing algorithms based on Marc_index.更多还原