I defend here the thesis of the independence between domain knowledge
and problem-solving knowledge, arguing against the dominance of the so
-called ''inter action problem'' mentioned in a recent paper by Van He
ijst, Schreiber and Wielinga to dispute the feasibility of a single do
main ontology shared by a number of different applications. The main p
oint is that reusability across multiple tasks or methods can and shou
ld be systematically pursued even when modelling knowledge related to
a single task or method. Under this view, I discuss how the principles
of formal ontology and ontological engineering can be used in the pra
ctice of knowledge engineering, focusing in particular on the interpla
y between general ontologies, method ontologies and application ontolo
gies, and on the role of ontologies in the knowledge engineering proce
ss. I will then stress the role of domain analysis, often absent in cu
rrent methodologies for the development of knowledge-based systems. (C
) 1997 Academic Press Limited.