Research And Implementation On Path Oriented Extensible Model For XML Database

As a popular standard for data on the Internet, XML usually has been accepted as the framework for semi-structured data. Several XML query and storage models have been proposed to store and retrieve XML data in Relation DBMS. Almost every of them involved in processing the Path Expression from different aspects, which are the core parts of XPath/XQuery. At present, there are two different approachs to process the path structure. One builds the path information into the index and converts the structural join into string match problem. But it will be confronted with the difficulty of complex Path Expression. Another approach decomposes the whole XML into sets of edge or node, which can be stored into the relation table. However, it will lead to a mass of relation join. In short, all the methods can't provide the extensible optimization mechanism.In this paper we present a new flexible model, called Path Oriented Extensible Model (POEM), which can quantify the stuctual information to fit all the complex problems in Path Expression. Unlike the existing solutions, it focuses on the path problem both the query and storage aspects simultaneously. It builds the store of coordinate mapping mechanism and query solution according to the key factor of the performance issues in the Path Expression evaluation – Structural Join. Then, it give the formal description of path query with Query Path Graph and the optimization strategy and argorithm. Then, we implement the XML Server with the middleware architecture according to the modle. At last, the performance evaluation and extensible ability analysis of XML Server are presented. It is proved that the extensible mechanism of XML Server based on POEM can support all kinds of requirement of XML application, and it exhibits the obvious advantage in performance.