Nowadays, scientific collaboration is prevalent in various scientific disciplines. Scientific collaboration has been resulted from knowledge complexity, increase in demand for more specialization, and interdisciplinary skills in research. It is a social phenomenon in research and has been studied systematically since the 1960s. Since then, some increase in the rate of scientific collaboration has been reported by various researchers.
Social network analysis is used for describing the scientific collaboration patterns identified by co-authorship relations (Stefano, Giordano & Vitale, 2011). Scientists included in the collaboration networks share their ideas, use similar methods and techniques for extracting and analyzing research data and influence each other's works. As one of the most documented and tangible forms of scientific collaboration and the most formal manifestation of intellectual share among authors in producing scientific works, co-authorship is the collaboration among two or more authors on producing a work that results in a production with higher quality and quantity than that produced by a single author (Hudson, 1996). Collections of such collaborations among researchers can construct a co-authorship network in which authors form nodes and the line between two nodes is considered as the co-authorship relation created in the papers. As a main category of social networks, the co-authorship network can be used for determining the structure of scientific collaboration and individual authors' research states (Liu et al., 2005).
On the other hand, one of the complex debates in bibliometrics is researchers' scientific influences. Since some authors relate a researcher's scientific influence to the citation rate of his/her works, scientific influence is not restricted to one's works and a researcher's interaction with other researchers in a field is at work when considering his/her scientific influence, i.e. his/her social influence. Social influence is one's ability to influence others by a means of social interaction processes (Truex et al. 2011). In other words, the expansion of a researcher's thoughts can be measured by studying his/her co-authorship trends in a certain scientific field (Cuellar et al. 2016). Three measures of centrality (degree, betwenness, and closeness) are often used for measuring the social influence. Centrality is one of the most important and common measures in analyzing social networks, especially for identifying main and powerful influencing actors.
Considering the above-mentioned points, this study aims at investigating the relationship between researchers' productivity and performance with their centrality measures among researchers in the iMetrics. Specifically, this study attempted to determine:
The rankings of iMetrics researchers based on their centrality (including degree, betweenness, and closeness) measures;
The possible relationship between productivity (the number of articles) and centrality measures; and
The possible relationship between performance (the number of citations) and centrality measures.
Several scholars have directly applied centrality measures to co-authorship networks in different fields (Barabasi et al. 2002; Otte & Rousseau, 2002; Mutschke, 2003; Liu et al, 2005, Acedo et al, 2006; Krichel & Bakkalbasi, 2006; Liu et al, 2007; Hou et al, 2008; Gomez et al, 2008). On the other hand, the study of research productivity, citation impact and collaboration has a long-standing tradition in LIS research, and these three indicators have been employed in many disciplines to measure research success in terms of output (Abrizah et al. 2014). To be more specific, the relationship between social network structures in co-authorship network and research productivity and impact is studied in several studies (Newman, 2001; Egghe et al. 2007; Abbasi and Jaafari 2013; Yin et al. 2006).
Among them, Hou, Kretschmer and Liu (2008) investigated the structure of scientific collaboration networks in scientometrics at the level of individuals by using bibliographic data of all papers published in the international journal Scientometrics during 1978-2004. The result showed that Glanzel is the central author of the whole network in terms of the highest degree, betweenness and closeness centralities, which indicates that he is the most influential person in the network. With respect to sub-networks. Moreover, they found a positive and significant correlation between output of authors and the centrality measures, which revealed that most of the prolific authors were also active in collaboration network in the field of scientometrics.
Yan and Ding (2009) indicated that co-authorship centrality measures are significantly associated with citation counts, with betweenness centrality having the strongest association. Badar et al. (2012) examined the association of co-authorship network centrality (degree, closeness and betweenness) and the academic research performance of chemistry researchers in Pakistan. Results related to regression revealed a positive impact of degree and closeness and negative impact of betweenness centrality on research performance. Temporal analysis using node-level regression confirmed the direction of causality and demonstrated a positive association of degree and closeness centrality on research performance.
Guns et al. (2010) found that top authors in Scientometrics and Journal of Informetrics had the highest global collaboration network centrality measures. Moreover, Liao and Yen (2012) indicated that the degree of research collaboration had a strong positive relationship with research productivity.
In a more recent study, Abrizah et al. (2014) investigated the field of informetrics to identify publication strategies that have been important for its successful researchers. They used a micro-analysis of informetrics researchers from 5,417 informetrics papers published in 7 core informetrics journals during 1948-2012. Findings revealed that the 30 most productive informetrics researchers of all time span several generations and seem to be usually the primary authors of their research, highly collaborative, affiliated with one institution at a time, and often affiliated with a few core European centres. Their research usually has a high total citation impact but not the highest citation impact per paper. Moreover, results indicated that the most cited authors also tend to be the most productive authors: 20 of the 30 most cited authors are also in the most productive 30. Based on betweenness centrality, Glanzel, Rosseau, and Leydesdorff gained the highest scores, respectively.
Results of Soheili, khademi and mansouri (2015) showed that there is a significant correlation between Journal Impact Factor (JIF) and all centrality measures except closeness centrality at P= 0.001. Results also showed that there is a significant correlation between productivity of authors and all centrality measures scores at P> 0.001. Also, regression reports direct relationship of degree, closeness and flow betweenness and inverse relationship of betweenness as well as Eigen vector centrality on productivity of researchers.
This research applied co-authorship analysis and social networking analysis. The research population consisted of the iMetric papers that were indexed in the Web of Science (WoS) during 1978-2014. It worth nothing that in research on fields such as bibliometrics, informetrics, webometrics...