XML Type Dependency and XML Type Normal Form

Abstract

As a theoretical foundation of relational database, integrity constraints such as functional dependency play very important role in the design of relational data schema. Moreover, the wide application of XML data requires relevant integrity constraints to normalize the design of XML documents and the design of XML database. Because XML data is a semi-structure data, we find that data redundancy in an XML document results not only from functional dependencies between the data values of different data items, but also from the dependencies between the structure of some polymorphic data and values of its members in XML Documents. As a new class of XML constraint, we propose the concept of XML type dependency (XTD), to indicate the type/value dependencies in XML documents. We give the formal definitions of the XML type dependency and its instance constraint and its satisfaction checking algorithms. We also propose a XTD-based normal form and a normalization algorithm for well-designed XML documents. We have also proved that the redundancy caused by the XTDs can be eliminated by this normalization.