The Impact of Storage Strategies on Maintenance of XML Views

Abstract: Information in a data warehouse is stored in materialized views, which must be kept consistent with respect to changes made in the sources. This problem has been extensively studied in the relational model. The process is referred to as view maintenance. XML is emerging as the de facto standard for data representation and data exchange of semistructured data. Most discussions involving XML assume the XML data is stored in plain text files. However, there are a number of different approaches for storing XML data, which can be categorized according to the underlying system used. Views and materialized views can also be specified in XML. This dissertation investigates how view maintenance in an XML context is influenced by the utilized approach for storage. We survey existing storage strategies using a relational database as the underlying system for storage, and storage strategies using plain text files. Further, we survey approaches for maintenance in the context of XML. We investigate three selected storage strategies in detail. We conclude with some insights gained during the investigation.