Assessment of the Performance of Fuzzy Cluster Analysis in the Classification of RFC Documents

The work described in this paper has been carried out in the context of a network-based teaching and learning system. In such systems, the need to find relevant learning resources to a particular knowledge domain arises. This need comes not only from a learning perspective, but also from an authoring perspective. From the authoring point of view, the re-use of documents to facilitate the delivery of new courses is seen as an important advantage. This helps to minimise the authoring task, as long as there is a way of locating relevant learning resources for the course in question. From the learning perspective, we must keep in mind that students have different profiles in terms of background knowledge, learning objectives, preferred learning styles, etc. Despite the fact that online courses might present a well-defined structure, which is conceived by the teacher, such structure should not be seen as a rigid track to follow, but as an orientation track. To cope with the different student profiles, the system should allow an adaptive navigation of the document space, based on a relevance measure of other documents to the ones that are part of the defined tracks. The question that arises is how to measure relevance and how to organise documents in an abstract knowledge space. To evaluate how the knowledge space could be formed, an experiment has been carried out with a collection of RFC documents. As a candidate technology, the use of fuzzy cluster analysis has been explored.