The non-transitivity of similarity relations is the main problem encountered when - as in taxonomic studies in biology - one wants to base classification schemes on observed similarities and dissimilarities. While recent advances in molecular biology give rise to new and rather abstract data structures which can easily be used as input for automatic classification procedures we are still very much in need of a better and deeper understanding of the many delicate points which need consideration once (semi-)automatic classification procedures are applied to biological or other data. The papers collected in this volume are devoted to precisely this problem. They study various theoretical aspects of three reconstruction methods in biology, and psychology, discuss their value in specific biological contexts, apply tree-like recursion networks in chess programming and indicate a conceptual framework for studying cluster analysis from a purely mathematical point of view.
