Even though XML was first introduced as a schema-less, self-describing data representation language, there are now proposals for XML schema descriptions. The addition of schema information opens new opportunities in using the already established database management technology for storing and handling XML data, since databases are traditionally focused on data that conform to a fixed, predefined schema. Schema information allows better data integrity, more effective data storage, and more efficient query evaluation. This paper describes an effective framework for storing XML data in an object-oriented database and an optimization framework for translating XML queries into efficient algorithms. We first present a new type system for describing XML data, well integrated with the ODL type system of the ODMG standard, that captures both schema-less (semi-structured) and schema-based XML data. We then introduce a small set of syntactic extensions to ODMG OQL, powerful enough to make OQL a full-fledged XML query language. Next, we present a framework for translating XML queries into OQL queries based on XML schema information. Instead of inventing yet another semi-structured algebra for expressing our translations, the target of our transformation rules is OQL code, which not only has precise semantics, but has also been the focus of various optimization techniques. Schema information is an indispensable component of our transformations. It is used in disambiguating terms with multiple interpretations such as wildcard tag projections, in choosing the storage format for the XML data, and in generating OQL code guided by the choice of storage.
展开▼
