| Summary: | Documents that comply with the XML standard are characterised by inherent ordering and their modelling usually takes the form of a tree. Nowadays, applications generate massive amounts of XML data, which requires accurate and efficient query-able XML database systems. XML querying depends on XML labelling in much the same way as relational databases rely on indexes. Document order and structural information are encoded by labelling schemes, thus facilitating their use by queries without having to access the original XML document. Dynamic XML data, data which changes, complicates the labelling scheme. As demonstrated by much research efforts, it is difficult to allocate unique labels to nodes in a dynamic XML tree so that all structural relationships between the nodes are encoded by the labels. Static XML documents are generally managed with labelling schemes that use simple labels. By contrast, dynamic labelling schemes have extra labelling costs and lower query performance to allow random updates irrespective of the document update frequency. Given that static and dynamic XML documents are often not clearly distinguished, a labelling scheme whose efficiency does not depend on updating frequency would be useful. The GroupBased labelling scheme proposed in this thesis is compatible with static as well as dynamic XML documents. In particular, this scheme has a high performance in processing dynamic XML data updates. What differentiates it from other dynamic labelling schemes is its uniform behaviour irrespective of whether the document is static or dynamic, ability to determine all structural relationships between nodes, and the improved query performance in both types of document. The advantages of the GroupBased scheme in comparison to earlier schemes are highlighted by the experiment results.
|
|---|
