Towards flexible learning object metadata

This paper outlines the research we are doing in acquiring, describing and using learning object metadata. Instead of the IEEE LOM and other standardised metadata schemes, we argue for a more flexible approach to both defining and associating metadata with learning objects. This approach, which we call the ecological approach, sees metadata as the process of reasoning over observed interactions of users with a learning object for a particular purpose. Central to this approach is the notion that Semantic Web enabled computational agents will both provide and consume pieces of actual usage data that have been collected about a learning object in determining the usefulness of this learning object for some new purpose. This is then an evolutionary approach to metadata creation as compared to move traditional prescriptive 'one size fits all' approaches.