Structure of hierarchic clusterings: implications for information retrieval and for multivariate data analysis

Research output: Contribution to journalArticle

17 Citations (Scopus)

Abstract

Hierarchic clustering methods may be used to condense information for a user, as they are in multivariate data analysis, or to achieve computational advantages, as they are in information retrieval. The structure of the hierarchic classification produced has a direct bearing on the effectiveness and utility of using cluster analysis, yet this important feature of the classification has only been implicitly referred to in the literature to date. In this study, three different coefficients are defined, each of which quantify the symmetry-asymmetry (balancedness-unbalancedness) of hierarchic clusterings on a scale from 0 to 1. Using examples of data from the areas of information retrieval and of multivariate data analysis, a number of hierarchic clustering methods are discussed in terms of the hierarchies they produce.

LanguageEnglish
Pages611-617
Number of pages7
JournalInformation Processing and Management
Volume20
Issue number5-6
DOIs
Publication statusPublished - 1 Jan 1984
Externally publishedYes

Fingerprint

Information retrieval
information retrieval
multivariate analysis
data analysis
Cluster analysis
cluster analysis
asymmetry
Clustering
literature

Cite this

@article{ecde5d5eb7c24316abc07e3b69975366,
title = "Structure of hierarchic clusterings: implications for information retrieval and for multivariate data analysis",
abstract = "Hierarchic clustering methods may be used to condense information for a user, as they are in multivariate data analysis, or to achieve computational advantages, as they are in information retrieval. The structure of the hierarchic classification produced has a direct bearing on the effectiveness and utility of using cluster analysis, yet this important feature of the classification has only been implicitly referred to in the literature to date. In this study, three different coefficients are defined, each of which quantify the symmetry-asymmetry (balancedness-unbalancedness) of hierarchic clusterings on a scale from 0 to 1. Using examples of data from the areas of information retrieval and of multivariate data analysis, a number of hierarchic clustering methods are discussed in terms of the hierarchies they produce.",
author = "F. Murtagh",
year = "1984",
month = "1",
day = "1",
doi = "10.1016/0306-4573(84)90077-3",
language = "English",
volume = "20",
pages = "611--617",
journal = "Information Processing and Management",
issn = "0306-4573",
publisher = "Elsevier Limited",
number = "5-6",

}

TY - JOUR

T1 - Structure of hierarchic clusterings

T2 - Information Processing and Management

AU - Murtagh, F.

PY - 1984/1/1

Y1 - 1984/1/1

N2 - Hierarchic clustering methods may be used to condense information for a user, as they are in multivariate data analysis, or to achieve computational advantages, as they are in information retrieval. The structure of the hierarchic classification produced has a direct bearing on the effectiveness and utility of using cluster analysis, yet this important feature of the classification has only been implicitly referred to in the literature to date. In this study, three different coefficients are defined, each of which quantify the symmetry-asymmetry (balancedness-unbalancedness) of hierarchic clusterings on a scale from 0 to 1. Using examples of data from the areas of information retrieval and of multivariate data analysis, a number of hierarchic clustering methods are discussed in terms of the hierarchies they produce.

AB - Hierarchic clustering methods may be used to condense information for a user, as they are in multivariate data analysis, or to achieve computational advantages, as they are in information retrieval. The structure of the hierarchic classification produced has a direct bearing on the effectiveness and utility of using cluster analysis, yet this important feature of the classification has only been implicitly referred to in the literature to date. In this study, three different coefficients are defined, each of which quantify the symmetry-asymmetry (balancedness-unbalancedness) of hierarchic clusterings on a scale from 0 to 1. Using examples of data from the areas of information retrieval and of multivariate data analysis, a number of hierarchic clustering methods are discussed in terms of the hierarchies they produce.

UR - http://www.scopus.com/inward/record.url?scp=0021600424&partnerID=8YFLogxK

U2 - 10.1016/0306-4573(84)90077-3

DO - 10.1016/0306-4573(84)90077-3

M3 - Article

VL - 20

SP - 611

EP - 617

JO - Information Processing and Management

JF - Information Processing and Management

SN - 0306-4573

IS - 5-6

ER -