Abstract

The eXtensible Markup Language (XML) has been adopted as the new standard for data exchange on the World Wide Web. As the rate of adoption increases, there is an ever pressing need to store, query and update XML in its native format, thereby eliminating the overhead of parsing and transforming XML in and out of various data formats. However, the hierarchical, ordered and semi-structured properties of the tree structure underlying the XML data model presents many challenges to updating XML. In particular, many of the tree labeling schemes were designed to solve a particular problem or provide a particular feature, often at the expense of other important features. In this dissertation, we identify the core properties that are representative of the desirable characteristics of a good dynamic labeling scheme for XML. We focus on four features central to the outstanding problems in existing dynamic labeling schemes; namely a compact label encoding, scalability, deleted node label reuse and a label storage scheme for binary-encoded bit-string node labels. At present there is no dynamic labeling scheme that integrates support for all four features. We present a novel compact and scalable adaptive encoding method to facilitate a highly constrained growth rate of label size under arbitrary node insertion and deletion scenarios and our encoding method can scale efficiently. We deploy our encoding method in two novel dynamic labeling schemes for XML that can completely avoid node relabeling, process frequently skewed insertions gracefully and reuse deleted node labels.