XML Semantic Query Optimisation

Abstract

XML Semantic Query Optimisation (XSQO) is a method that optimises execution of queries based on semantic constraints, which are extracted from XML documents. Currently most research into XSQO concentrates on optimisation based on structural constraints in the XML documents. Research, which optimises XML query execution based on semantic constraints, has been limited because of the flexibility of XML. In this thesis, we introduce a method, which optimises XML query execution based on the constraints on the content of XML documents. In our method, elements are analysed and classified based on the distribution of values of sub-elements. Information about the classification is extracted and represented in OWL, which is stored in the database together with the XML document. The user input XML query is evaluated and transformed to a new query, which will execute faster and return exactly the same results, based on the element classification information. There are three kinds of transformation that may be carried out in our method: Elimination, which blocks the non-result queries, Reduction, which simplifies the query conditions by removing redundant conditions, and Introduction, which reduces the search area by introducing a new query condition. Two engines are designed and built for the research. The data analysis engine is designed to analyse the XML documents and classify the specified elements. The query transformation engine evaluates the input XML queries and carries out the query transformation automatically based on the classification information. A case study has been carried out with the data analysis engine and we carried out a series of experiments with the query transformation engine. The results show that: a. XML documents can be analysed and elements can be classified using our method, and the classification results satisfy the requirement of XML query transformation. b. content based XML query transformation can improve XML query execution performance by about 20% to 30%. In this thesis, we also introduce a data generator, which is designed and built to support the research. With this generator, users can build semantic information into the XML dataset with specified structure, size and selectivity. A case study with the generator shows that the generator satisfies the requirements of content-based XSQO research.