ADAPTIVE CLUSTERING OF HYPERMEDIA DOCUMENTS

A hypermedia system connects various types of information into a netwo rk where related nodes of information (text, audio, video) are connect ed by links. Clustering these nodes is an effective way to reduce info rmation-overhead, allowing the user to browse through the clusters as well as the individual nodes. In this paper, we compare the use of two adaptive algorithms (genetic algorithms, and neural networks) in clus tering hypermedia documents. These clusters allow the user to index in to this overwhelming number of nodes and find needed information quick ly. We base the clustering on the user's paths through the hypermedia document and not on the content of the nodes or the structure of the l inks in the document, thus the clustering reflects the unique relation ships each user sees among the nodes. The original hypermedia document remains untouched, however each user will now have a personalized ind ex into this document. Copyright (C) 1996 Elsevier Science Ltd