Hierarchical author networks: An analysis of European molecular biology laboratory (EMBL) publications

Abstract

Co-authorship analyses are both difficult to perform and interpret. We have devised a new way of calculating and representing hierarchical author networks that depict relationships among authors in a more exhaustive and less equivocal manner than most available automatic analyses. Any structure, however complex, can be broken down into independent subclusters of authors that can be represented as individual interconnected networks. We illustrate our approach by analysing the authors of publications giving the European Molecular Biology Laboratory (EMBL) as an affiliation in 1994 (from the ISI 1994 CD-ROM). The networks can be interpreted by referring to the official EMBL staff list (Annual Report 1993) and, in terms of research topics, by consulting the article titles and abstracts. In this respect, correspondence analyses of the author-publication matrices—that are the counterparts of the author-author matrices—prove extremely useful in structuring the thematic information. In fact, both methods—the hierarchical author networks and the correspondence analysis biplots—mutually enrich each other and provide a global picture of the inherent structure and interests of the EMBL as given by their 1994 publications.