Extended and Multiple Tree Representations of Non-Hierarchical Clustering Solutions

Non-hierarchical clusterings of proximity data can represent a more general class of solutions than simple tree structures: those solutions consisting of overlapping (non-nested) sets of objects. However, graphical r representations of such non-hierarchical clusterings are usually quite complex and difficult to read. Carroll and Pruzansky (1972) suggested representing non-hierarchical solutions with multiple ultrametric or additive trees. Corter and Tversky (1986) introduced the extended tree (EXTREE) model, which represents a non-hierarchical distinctive-feature structure as a tree plus a small number of overlapping clusters.