Network Models for Scaling Proximity Data
R. Duncan Luce, Donald D. Hoffman, Michael D'Zmura, Geoffrey Iverson, A. Kimball Romney · Psychology Press eBooks · 2013
The analysis of similarity and dissimilarity data has focused on two main approaches, the spatial and the discrete approach. In spatial models such as multidimensional scaling (Shepard, 1962; Kruskal, 1964), dissimilarities are represented by means of a continuous, low-dimensional space. Other methods derive from discrete models that yield hierarchical clusters represented as ultrametric trees (Johnson, 1967; Hartigan , 1967), more general tree structures and multiple tree structures (Carroll & Chang, 1973 ; Carroll , 1976; Sattath & Tversky, 1977; Cunningham, 1978; Carroll & Pruzansky, 1980; De Soete , 1983; Carroll, Clark, & DeSarbo, 1984), structures called extended trees (Corter & Tversky, 1986), additive clusters (Shepard & Arabie, 1979), and network models (Feger & Bien , 320 KLAUER AND CARROLL 1982; Feger & Droge, 1984; Orth, 1988; Schvaneveldt, Dearholt, & Durso, 1988; Hutchinson, 1989; Klauer, 1989; Klauer & Carroll, 1989, 1991).