The recognition of kernel research team. Issue 4 (November 2022)

• Record Type: Journal Article
• Authors: Ma, Guoshuai; Yuhua, Qian; Zhang, Yayu; Yan, Hongren; Cheng, Honghong; Hu, Zhiguo
• Is Part Of: Journal of informetrics. Volume 16:Issue 4(2022)
• Publication Date: 2022-11
• Subjects: Kernel research team -- Co-authorship network -- Core members; 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.2022.101339
• Languages: English
• ISSNs: 1751-1577

Abstract

Highlights: The definition of local mutual degree of dependence : Inspired by Adams's fourth elements and Axiomatic index, a local index is proposed to measure the relationships between any pair of co-authors. In contrast to node filtration through the fixed threshold, our algorithm automatically distinguishes the kernel authors; the iteration converges quickly; the output network contains a small number of authors who have published plenty of papers and get much attention. The presence of Pareto principle in science : The statistics of co-author networks from ten research fields show that a small number of authors published the most papers and achieved the most citations in the networks, indicating the presence of the Pareto principle in co-authorship networks. Kernel research team recognized : A fast attributed overlapping community detection method is applied to the academic networks for detecting kernel research teams. The degree distributions of these kernel teams are uniform flat than that of the communities detected by original co-authorship networks. Abstract: Scientific projects are usually created by teams rather than individuals since the realizations of the projects need complex instruments and multidisciplinary cooperations. Although there is a myriad of reports on the assembly mechanisms of research teams, most are restricted to the empirical analysis of some special teams, and they failed to analyze the research team from big co-authorship networks. Inspired by L. G. Adams's "basic elements" of the successful research team, this paper proposed a method for identifying the kernel research teams from the co-author networks. We create a database containing all articles published in the journals recommended by the China Computer Federation (CCF), based on which the networks of ten subfields in computer science are constructed. In the empirical analysis, a handful of scholars are found to contribute a large portion of literature and gather numerous citations; this proves the presence of the Pareto principle in academic networks. Furthermore, the information of 34 famous research teams is collected and analyzed; our study shows most leaders and members who belong to these 34 teams can be recovered from the network of kernel research teams even when more than 70 % of authors are removed from the original co-authorship network. Finally, in order to take full advantage of the authors' research interests, we improve the original label propagation method to guarantee good performance in our dataset.