RelAndXML: a system to manage XML-based course material with object-relational databases

In this thesis, we present our newly invented system RelAndXML for the management and storage of hypertext-centric XML documents and the according XSL stylesheets. Our sample application area is the course material at university. Typically, course material is being reused on multiple assignments, while it is also important to add or replace questions. Currently, teaching assistants use different word processors and the availability of previous year’s assignment files is not always certain. This results sometimes in retyping the same questions. RelAndXML offers a solution to this problem by saving the XML formatted assignments as text modules and other parts in a special database. However, RelAndXML is not restricted to this application area, since the core schema can easily be replaced, e.g. by a schema for technical online manuals. Storing XML documents in databases has been a major topic in database research in the last few years. Approaches on this topic are dependent on the desired application area and can be divided into two main directions: some concentrate on data-centric and others on document-centric documents. Data-centric XML documents are valid with respect to a document type definition (DTD), they have sparse mixed content and the order in which subelements and PCDATA occur is usually unimportant. The concept for storing data-centric XML in (object-) relational database systems is to define a mapping between the DTD and the database schema. Since our hypertext does not conform to a DTD, we cannot use a data-centric approach on its own. Document-centric XML often has no DTD such that it is non-valid XML, mixed content often occurs, and the document order does matter. The known document-centric approaches do preserve document order, but they make reusing text modules impossible. In this thesis, we present a hypertext-centric approach that combines aspects of datacentric as well as document-centric approaches. It enables the reuse of text modules and preserves document order where necessary. Another important feature of RelAndXML is that it supports not only XML elements, but also attributes, comments, and processing instructions. RelAndXML is a Java-implemented system using an object-relational database. It provides a graphical user interface which allows to create and update the XML and XSL parts of hypertext documents, to insert new or saved text modules, and to generate HTML or XHTML output documents.