Learning object repositories with dynamically reconfigurable metadata schemata

In this paper we describe a model for learning object repositories in which users have full control over metadata schemata. Thus, they can define new schemata and reconfigure existing ones in a collaborative fashion. In consequence, the repository must react to changes in schemata in a dynamic and responsive way. Since schemata enable operations like navigation and search, dynamic reconfigurability requires clever indexing strategies, resistant to changes in these schemata. For this purpose, we have used conventional inverted indexing approaches and have also devised a hierarchical clustering-based indexing model. By using Clavy, a system for managing learning object repositories in the field of the Humanities, we provide some experimental results that show how the hierarchical clustering-based model can outperform the more conventional inverted index-based solutions.