Domain analysis has been recognized to be an important factor in systems engineering. However, due to the relatively short time we have
spent investigating this concept of a conceptual model on a level above
the single project, we still face severe problems with respect to
methodologies and tool-support. Our contribution focuses on the problem
to decide on similarities between (data) objects. We suggest the use of
a meta (data) structure which handles similarity as a semantic concept.
This model goes beyond the usual classification and generalization
techniques by supporting so called "generic properties". Using such a
meta model, we will be able to define an algorithm which retrieves from
the meta data structure a set of objects which are similar to a user-defined starting set of object descriptions. This approach will provide
us with a basis to integrate objects of an application domain into one
comprehensive structure and to approach the problem of multiple overlapping domain models.