چکیده:
In Bibliometric and scientometric studies, three approaches or” multivariate analysis, namely Factor A nalysis, Cluster Analysis and Multidimensional Scaling are the most used methods. 1”liis article aims to review the use and applications of these techniques.
خلاصه ماشینی:
ir In Bibliometric and scientometric studies, three approaches or” multivariate analysis, namely Factor A nalysis, Cluster Analysis and Multidimensional Scaling are the most used methods.
Generally, three approaches to multivariate analysis have been used in citation studies (including bibllometric and scientometric studies to display the inter-documentany relationships in the matrices ol'the most highly cited/co- cited documents) as follows: Cluster Analysis, Multidimensional Scaling and Factor Analysis.
MULTIVARIATE ANALYSIS The branch of statistics dealing with procedures for summarizing, representing and analyzing multiple quantitative measurements of a number of individuals or objects is called ‘Multivariate Analysis’ [2].
Experimentally, ‘clusters’ or groups of highly similar entities are formed by these procedure* l*1- Although Cluster Analysis has been recognized in the current century, most of its literature has been provided during the past two decades [8].
Iranian Journal of Information defence & Technology, Volume 1, Number 2 July / December, 2003 The Statistical Packagc for the Social Sciences (SPSS), for Windows, provides a clustering program that implcmcnts seven (i.
As in most citation analysis studies [11,23,25,26], this article used a two-dimensional solution of MDS (ALSCAL in SPSS) to assign relativc locations to clusters based on thc ranked order of their citation linkages [l l,23,26].
For example, for this study, two facto> are extracted by using Principal Component method: Factor 1 that accounts for thc largest portion of variance (75.
Especially, by using three methods of dimensionality reduction, namely Cluster Analysis, Multidimcnsional Scaling and Factor Analysis in bibliometric and scientometric studies, large and irrelcvant data will be reduced to a few significant and interpretable groups or dimensions.