Repositories of educational content enable a worldwide exchange of learning resources and support an economic use of existing materials. Metadata provide information describing objects and are essential for the exchange and re-use of content. Due to a rapidly growing amount of content, the scalability of metadata creation is becoming an important issue. The creation of structured metadata by experts is cost-intensive and does not scale. In contrast, automatically generated metadata are cheap to obtain, but provide only limited information on the information content of objects. A third way of enriching objects with metadata is social tagging, resulting in folksonomies, i.e., collaboratively user-generated, unstructured vocabularies. Combining these three approaches can provide for a cost-effective way of enriching content with useful types of metadata allowing users to find resources that fit their needs. This approach is pursued in the MELT project…