This paper describes an environment that supports novice information system
s designers in initial concept formation by oganising design 'cases' in evo
lving networks, automatically clustered by a neural agent to suggest genera
lisations of design patterns. Flexible reframing of the cases also supports
creativity by uncovering less obvious but relevant design concepts. If the
responsibility of linking related design cases is distributed to a populat
ion of learners who contribute their cases and tentative generalisations, g
enetic techniques can be applied to sustain the evolution of the base of ca
ses towards ontological forms that better guide and constrain the case reus
e process, thus supporting knowledge transfer. (C) 1998 Elsevier Science B.
V. All rights reserved.