An analytical system for evaluating academia units based on metrics provided by academic social network

Abstract Social networks are becoming more and more popular, not only among young people looking for entertainment, but also among specialists, experts and researchers who wish to establish professional networks, develop business or research projects. They may be useful also for the comparison and evaluation of scientists and research organizations. This study aims to show how to build a framework of an analytical system for evaluation of researchers and research units using the data retrieved from an academic social network. Acquired data are used to find out the main differences between ResarchGate (RG) usage and values of metrics owned by scientists of different gender, scientific title and field of study to find out if various groups of employees can be directly compared. The authors apply web scraping technique for collecting data from university web page (2847 employees) and use R scripts to acquire the metrics form RG portal. Also, data of 1497 researchers and teaching workers from 11 faculties at Polish university were explored. The descriptive statistics, Chi square test, ANOVA and logistic regression were used to analyse the main RG metrics: RG Score, number of publications, reads and citations. Analysis shows the significant differences both in terms of popularity of ResearchGate and values of its main metrics. The research confirmed that 1) the rvest package allows for fast data acquisition from RG, 2) RG metrics can be used by university managers to compare achievements and progress of single researchers, research labs, departments or faculties, 3) Researchers employed at the faculties of formal and natural sciences use RG portal more frequently, possess higher values of RG metrics, therefore different types of workers and various branches of science shouldn’t be compared directly.