Premium
CHARACTERS AND CLUSTERING IN TAXONOMY: A SYNTHESIS OF TWO TAXIMETRIC PROCEDURES
Author(s) -
Legendre Pierre,
Rogers David J.
Publication year - 1972
Publication title -
taxon
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.819
H-Index - 81
eISSN - 1996-8175
pISSN - 0040-0262
DOI - 10.2307/1219157
Subject(s) - cluster analysis , computer science , character (mathematics) , hierarchical clustering , graph , taxonomy (biology) , natural language processing , artificial intelligence , interpretation (philosophy) , data mining , theoretical computer science , information retrieval , mathematics , biology , botany , geometry , programming language
Summary The problem of producing a classification from data gathered on specimens has two main components: first the information about the specimens must be structured as characters and character states in such a way that it carries the most information about the taxonomic structure of the objects under study, the mathematical ‘noise’ being eliminated as much as possible. Then this information must be handled in such a way that a hierarchical partitioning of the objects, called classification, is derived. This paper presents computer‐aided methods for the accomplishment of these steps. These methods were worked out to be both mathematically and biologically sound. The character analysis method (called CHARANAL) uses information theory to measure the amount of information common to pairs of characters, and derives from it various measures for the comparison of characters. The clustering technique presented here (entitled GRAPH), on the other hand, is based upon graph theory, and is intended to represent the thought process of the ‘classical’ taxonomist. For each method are given a general explanation, a detailed explanation of the mathematics involved, an example, and a section on interpretation of results.