The paper points to weaknesses of conceptual modelling languages that
are oriented towards certain aspects of the problem domain of informat
ion systems (IS) development, e.g., activities, information resources,
objects, or actors. It is concluded that modelling languages are need
ed that allow modellers to 1) freely choose to represent a wide and ex
tensible range of aspects of problem domain phenomena contingent on th
e problems at hand; 2) simultaneously co-represent several aspects of
the same problem domain phenomenon whenever needed; 3) reflect semanti
cal relations between these aspects in the problem domain models; and
4) extend the set of kinds of aspects that can be represented and visu
alised throughout problem analysis as understanding of the problem dom
ain and the problems at hand increases. An approach called facet model
ling of real-world problem domains is therefore outlined to deal with
the complexity of contemporary analysis problems. It is shown how face
t models can be defined and visualised, before facet modelling is disc
ussed in relation to other recent ideas and techniques in the IS devel
opment field. Case studies are currently in progress to evaluate vario
us implications of the facet modelling approach empirically. (C) 1997
Elsevier Science Ltd.