After XML was proposed as the standard for representing and exchanging data on the World Wide Web, a large amount of XML data has been stored in DBMSs and ORDBMS is widely used as an XML repository. The structural constraints of a XML document are defined in a schema, like DTD and XML Schema, and XML Schema has been considered to be superior to DTD in terms of various facilities. Since there are applications requiring frequent updates of XML data, an efficient updating method of XML data is requried. It is necessary to provide updates conforming to XML Schema to produce XML documents, which are valid with respect to XML Schema. However, few systems guarantee the XML Schema-validity of data after an update. In this thesis, we propose an update method for XML data in an ORDBMS and an efficient method to verify the XML Schema-validity for updated XML data.
First, we propose a method for matching a data node in an XML document to its corresponding schema component in an XML Schema. Second, we analyze the XML Schema-validity constraints which depend on the type of node in an XML document, and classify the required data states and the schema information for verifying the validity of an update. Then, we design a method for verifying the XML Schema-validity. In addition, we design an update algorithm for each type of update on XML data in an ORDBMS. Third, we analyze the structure of XML Schema and propose a storage method for XML Schema, named as‘the storage method by type’, in order to efficiently extract the schema information for validity-verification. Finally, we provide the performance results of our proposed update algorithms. We show that ‘the storage method by type’is an efficient storage method for XML Schema by comparing the extraction times of schema information that has been stored using ‘the storage method by type’ with those using a naive method.
