Research On The Technique Of Schema-Based XQuery Optimization And Parallel Processing

With the repaid development of Web techniques and Web applications, XML is emerging as a standard data format for data presentation and exchange on the Web. It is widely applied in many important areas, such as electronic data exchange, science data, e-commerce and search engines, and there are a large number of XML documents existing on the Web. However, XML data is a kind of semi-structured data which has features from both text and structured data, which takes challenges on querying XML data. Though a language XQuery has been proposed by W3C organization to satisfy the requirement of querying XML data, the feature of semi-structured data and the complexity of XQuery make that traditional technique of query optimization in database is hardly satisfied for the performance requirement of XQuery processing.To improve the efficiency of query processing on XML data, this paper mainly researches on the technique of query optimization for XQuery and the technique of parallel processing on XML data. For query optimization, this paper proposes a kind of schema-based XML index, SBDI, which supports dynamic update and use paths in XML schema to improve the query efficiency of path expression in XQuery. Firstly, XML Schema firstly is mapped as a tree. We encode both XML Schema and XML document with the DDE encoding. The document index and schema index is created based on the encoding information, and the schema encoding information is also integrated into the document index. The advantage of this index is there is no encode vacuum reserved for update, which helps the index supports data update without reconstructed. Furthermore, this paper proposes the strategy of query processing based on SBDI index, which includes two optimizing strategies for processing path expression and three rewriting rules for FLWOR expression.To address the efficiency problem of querying on large XML document, this paper researches on the technique of parallel processing in Native XML Database System based on the architecture of pureXBase system. By analyzing the parallel processing mechanism for XML data and PSPIB parallel partitioning strategies, based on XMark benchmark, this paper proposes two strategies of query decomposition for path with "//" expression and FLWOR expression, and three stra(?)egies of results merg(?)g for eliminating redundancies, type conversion and sequence scan.The experimental results show that the index and optimizing strategies we proposed has higher performance on query processing than existing methods, and the cost of update operation is less than existing index structure. The experiments on parallel processing verified that strategies of query decomposition and strategies of results merging can improve the performance of processing large XML documents on processing time and speedup ratio. The research work of this paper will be helpful to improve the performance of processing XML data in database system.