New homepages: 
at Old Dominion University
and at Los Alamos National Laboratory!
Deze pagina is ook beschikbaar in het Nederlands
Old homepage (no longer up-to-date)
Learning hypertext networks, associative and distributed knowledge representation, dynamical and evolutionary approaches to knowledge development
I'm currently investigating learning rules that use the paths users follow through a hypertext to re-organise the network, so that it restructures itself into a representation of the users shared semantic structure.
The assumption that led to the development of hypertext/media interfaces is that hypertext networks, because they use associative and distributed representations of knowledge, are in some way more compatible with how humans store and retrieve information. Networks that more accurately represent their users' semantical structure will thus allow users to retrieve information more effectively.
As most hypertext networks, such as the WWW, are at present developped and designed by human web masters, that have only very limited knowledge and insight into what a good hyper structure should be, it can be expected that information retrieval and storage is at present highly inefficient. Our research suggests a number of locally learning rules that can automatically, without human intervention, make hypertext networks re-organise and streamline their structure
This link points to a short paper titled Adaptive Hypertext Networks That Learn The Common Semantics Of Their Users., which will be published in the proceedings of the International Congres on Cybernetics in Namur, August 21-25, 1995. You are invited to read this article and mail reactions, criticism or suggestions via the above provided E-mail link.
Photo: giving a seminar on learning webs at Goddard Space Flight Center, NASA.
