A Comparison And Evaluation Of Graph Theoretical Clustering Technioues
Derek Gordon Corneil, M.E. Woodward · INFOR Information Systems and Operational Research · 1978
In this paper we develop a methodology for analysing and comparing graph theoretical clustering algorithms. In order to select particular representative methods for a sample comparison, graph theoretical clustering methods are classified as clique, connectivity, and minimal spanning tree methods. The particular representatives of these categories are the Gotlieb-Kumar, Vaswani, and Zahn methods, respectively. The comparison of such algorithms requires a standard for generating statistically valid data as well as determining appropriate measures of the "goodness" of the clusterings.