abstract:
  Most of today's theory and empirical work cannot serve as a
  foundation for designing or applying evolutionary algorithms in
  novel areas of application. This article proposes qualitative models of
  interactions in evolutionary algorithms as a means to close that
  gap. It presents a knowledge representation in the form of a
  hypergraph and exemplifies the integration of empirical results into
  the model by an examination of contracting recombination operators
  with respect to the diversity. Both the use of a big unifying model and
  small focused models are discussed. In the long term, this model is aimed
  to support the design, application, and debugging of evolutionary algorithms.