| As a data exchange and representation standard, XML (eXtensible Markup Language) is being employed in a wide range of industry sectors, for use in product catalogues, scientific data repositories, and electronic office filings. With the increasing production and dissemination of information in XML, how to efficiently search large amounts of XML documents has become an important problem.; After decades of research and development, relational database technology is now mature. Relational database systems are readily available and widely used for storing and processing data. The emergence of XML documents presents exciting new opportunities as well as interesting challenges to relational database systems.; In this dissertation, we tackle the problem of querying XML documents by an approach called XML indexing. An XML index consists of a collection of structured labels. We encode XML documents to generate such an index, and process path expression queries using the structured labels in the index.; The efficiency of an XML indexing method depends on intrinsic properties of the structured labels, the processing algorithm on the labels, and the efficiency of the underlying database system. We explore this efficiency issue and show that current relational database systems have strengths but also limitations in supporting XML indexing workloads. We seek reasons for these strengths and limitations, and provide possible solutions to eliminate some of the limitations. |
