The British Broadcasting Corporation’s “Content Management Culture” department is rolling out a Content Management System across bbc.co.uk sites, adhering to structured content management principles; notably atomised content described with controlled vocabularies.

To create a new site in our CMS, information architects analyse the content on the site and create a Content Object Model (COM). The COM is then implemented in the CMS through XML configuration files that define the content creation and editing interface.

In order to enable to maximum degree of content sharing and re-use both within and outside of the BBC, we have undertaken to share as much of our content models as possible between departments. This creates a management problem, as keeping track of the differences between content models, and creating new models based on existing ones, is difficult to do with simple schemas or documentation.

To solve this problem we created a “metamodel” that describes our content modelling structure, and set about finding a tool that would let us create and compare models according to various features. After evaluating and discarding UML and other standard modelling tools, we decided Protégé’s flexible RDF/OWL-based modelling environment suited our needs, and its Java-based extension framework allowed us to create plugins that automatically generate our XML templates and interface definition files.

This paper discusses how RDF and OWL helped us to model our system, how we used object constraint language constructs to define our metamodel, how the open-world model of RDF can make life difficult for tool developers, whether the use of RDF reasoners is appropriate for these tools, and more.