Approximative Retrieval of Attribute Dependent Generalized Cases

In our current research project about electronic designs the problem to retrieve attribute dependent generalized cases occurs. This kind of cases cover an arbitrary subspace rather than a point in the space, spanned by the cases attributes and by a set of constraints. For such representations, the similarity assessment between a point query and generalized cases is a difficult problem for which this paper presents a solution. We developed and evaluated a converter, which samples the spawn subspace for each case, selects reasonable point cases out of it, and creates a new case base with only point cases belonging to the current generalized case. For this case base well-known retrieval engines can be used where only less modifications have to be made.