A geometric graph model for coauthorship networks. Issue 1 (February 2016)

• Record Type: Journal Article
• Authors: Xie, Zheng; Ouyang, Zhenzheng; Li, Jianping
• Is Part Of: Journal of informetrics. Volume 10:Issue 1(2016:Jan.)
• Page Start: 299
• Page End: 311
• Publication Date: 2016-02
• Subjects: Coauthorship network -- Modelling -- Geometric graph -- Hypergraph; Library statistics -- Periodicals; Information science -- Statistical methods -- Periodicals; Bibliometrics -- Periodicals; Bibliothèques -- Statistiques -- Périodiques; Sciences de l'information -- Méthodes statistiques -- Périodiques; Bibliométrie -- Périodiques; 020.727
• DOI: 10.1016/j.joi.2016.02.001
• Languages: English
• ISSNs: 1751-1577

Abstract

Abstract : Highlights: A geometric graph is proposed to illustrate certain factors engendering collaborations, such as the relativity of research interests, academic influences of authors, etc. The model gives a geometric view to understand collaborations at the level of research teams instead of authors, and illustrates the scale-free property of coauthorship networks as a consequence arising from the inhomogeneous sizes of research teams and article teams. The model also successfully predicts some other statistical features of empirical coauthorship networks, such as degree assortativity, high clustering, giant component, etc. Abstract: Modeling coauthorship networks helps to understand the emergence and propagation of thoughts in academic society. A random geometric graph is proposed to model coauthorship networks, the connection mechanism of which expresses the effects of the academic influences and homophily of authors, and the collaborations between research teams. Our analysis reveals that the modeled networks have a range of features of empirical coauthorship networks, namely, the degree distribution made up of a mixture Poisson distribution with a power-law tail, clear community structure, small-world, high clustering, and degree assortativity. Moreover, the underlying formulae of the tail and forepart of the degree distribution, and the tail of the scaling relation between local clustering coefficient and degree are derived for the modeled networks, and are also applicable to the empirical networks.