Abstract
Although many sophisticated CAD integrated environments have been developed and are currently adopted by enterprises in the design of mechanical parts and components, such kinds of tools should be extended in order to reach higher levels of performance. To this aim, Artificial Intelligence techniques are particularly suitable to provide CAD tools with a sort of "intelligence" typical of human experts. In particular a complex mechanical object to be designed exploiting CAD systems can be considered as an aggregation of simpler components that have to be put together in order to satisfy precise design rules owned by expert designers core knowledge. This paper presents an ontological approach to the problem of representing relationships among step-by-step more complex parts, in order to obtain a final product that fully meets initial requirements.