INTEGRATING UNSUPERVISED DOCUMENT CLASSIFICATION WITH A FOLKSONOMY IN A REPOSITORY OF LEARNING OBJECTS

摘要

In this paper, we describe a repository of learning objects which automatically organizes the content in clusters of semantic proximity by the use of text mining mechanisms and a social classification system. The automatic classification system takes advantage of an integration between the repository and an educational information system, which dynamically produces metadata. This unsupervised classification is enhanced with a folksonomy which provides an additional and non-automatic characterization of documents. We describe the initial classification process to create clusters of documents and the methodology sustain this organization during ingestion of new learning objects and new metadata.