A structured family of clustering and tree construction methods (Q1604343)

In this article, the work of \textit{V. Moulton} and \textit{M. Steel} [Discrete Appl. Math. 91, 215-233 (1999; Zbl 0914.05016)] is extended. The authors generalize Apresjan's clustering and Buneman's splitting methods to two families of tree-based classification methods by relaxing the definitions of isolation and separation indexes, respectively, and allowing weighting functions to vary. The classes of clusters and splits constructed (strong clusters, clean clusters, stable clusters, Buneman splits, refined Buneman splits, clean splits, and stable splits) are studied in detail. Among others, links among the methods obtained are established and complexity results for the construction of the classes of clusters and splits are given.