Study On The Parallel Query Based On The Index For A Native XML Database

Because of the expansibility, structure, platform independence and self-description, XML is now a proposed standard for web information publishing and information exchanging. But currently, RDBMS couldn’t support XML very well. Native-XML databases are constructed with extensive applications of XML. They are specialized for storing and managing XML data. Parallel processing provides an effective solution for large XML documents. How to improve system performance has been largely concentrated on optimizing XML queries in the parallel environment.Firstly, this thesis analyzes the character of XML documents after partitioning and proposes the strategy of partationing the query plan based on the single path. We design an effective XQuery algebra system for the parallel query. This query algebra is based on the Set-Oriented strategy. The system contains nine operators. This thesis also proposes some optimization strategies based on the algebra system. Secondly, this thesis studies the current XML indexing strategies. For the three kinds of instance in the single path, the corresponding indexes are designed. And based on the character of the index, structure joining algorithm is proposed to accelerate the path query. Finally, this thesis analyzes the XML keyword query and studies the keyword query based on SLCA (Smallest lower common ancestor) in parallel XML databases. We propose two efficient algorithms-SONB and MSOP. SONB (Scan once with no buffer) is used in the single machine and MSOP (Merge strategy based on ordered partition) could compute the SLCA efficiently in the parallel environment.Lots of experiments are performed to evaluate and analyze the solutions. The experimental results show that the XQuery algebra system has high efficient query performance in the parallel environment. The index plays an important role on the single path query. Our proposed approach achieves high efficiency for the keyword search.