Abstract
In the field of case-based reasoning (CBR), the derivation of so-called ‘similarity measures’ is an unresolved open question. In this work dimensional analysis is used to derive appropriate similarity conditions for a CBR technique. For the subclass of all case descriptions consisting of real-valued quantities with physical units, it is shown how the Pi-Theorem can be used to construct similarity conditions from these case descriptions. Within this approach a proof for the correctness of the CBR technique can be derived.
The properties of the CBR technique using dimensionless groups indicate that it bears some potential in engineering design, where knowledge in the form of analytical equations is not always available. Often only pointwise and/or incomplete knowledge about the future design object in the form of previous designs, prototypes or simulation results is available and appears in certain cases to be sufficient for the new CBR technique.
