Occurrences

Paper

../papers/03-02-03/03-02-03.html

Date of Presentation

Wednesday, 22 May

Time of Presentation

11.00

Presentation Level

In-The-Middle

Abstract

Databases provide persistent storage of structured data and tools to access and manipulate those data rapidly. XML provides a standard way to exchange structured information and transform form one representation to another. For many years, quite naturally, there has been much technical work on the linkage of XML and databases. That work has intensified in the last year, and polarized to some extent into two camps. On one hand are technologists who are concentrating on the flow of structured information between XML and relational databases -- packing XML to the database for persistent storage, reconstructing it as XML for data processing and exchange. In the other camp are technologists creating a new breed of native XML databases, in which XML data structures are mapped directly to disk for persistent storage. To my mind, there are in fact three types of 'XML database' -- ie three different ways in which structured data manipulated as XML in an application can be stored persistently in a database. The first type is an XML generating database, in which data are stored according to some database schema, with XML structures generated from some packing and unpacking process which maps between the database schema and the XML application schema. The second type is an XML document database, where the database acts as a document repository, managing the import, export, check-out, check-in and versioning of XML documents that are held as self-contained entities in the database schema. The third type is the XML component database where the database schema is the same as the XML schema and there is a one-to-one mapping between objects in the database and objects in the XML application -- this type of database has existed for many years built upon object-oriented technology and has more recently been implemented using 'native' XML database technology. These types of database are interesting to the theoretician, but provide a dilemma for the applications designer in the real world. Should an application be designed from a document-centric perspective, using an XML DTD or schema as the data model, or from an information-centric perspective, using a UML or database schema as the data model. This presentation shows how a simple application for producing, managing and distributing news articles can be designed using either an XML document model or a UML/relational database model. The advantages and disadvantages of each approach are discussed, to give some insight into how the right tools for the job can be chosen.