Grouping of semistructured data for efficient query processing

• Main Author: Neumüller, Mathias
• Format: Dissertation
• Language: English
• Published: University of Strathclyde 2004
• DOI: 10.48730/ee04-wa56

With the emergence of large-scale distributed computing applications semistructured data models have gained significant importance. Current practical semistructured data management systems can often not provide the performance required by practical applications. This work describes a model for the optimisation of semistructured data processing based on data groupings. Such groupings are of fundamental importance for efficient querying of semistructured data. The semistructured model does not imply the natural organisation of data that characterises rigidly structured representations. Instead, data groupings in the semistructured case must be derived from the data itself or its applications. This thesis presents a number of such possible data groupings and formalises them into a concept of domains. Different classes of domains are identified and the impact on different data sources is evaluated. A particular definition is then used to implement an efficient physical representation using an approach based on dictionary compression adapted from relational data management. Finally this approach is combined with a data grouping aimed at the efficient resolution of structural constraints.