Scholarly publications are among the most tangible forms of knowledge production. Therefore, it is important to analyse them, amongst other features, for gender or country differences and the incumbent inequalities. While there are many quantitative studies of publication activities and success in terms of publication numbers and citation counts, a more content-related understanding of differences in the choice of research topics is rare. The present paper suggests an innovative method of using term communities in co-occurrence networks for detecting and evaluating the gender- and country-specific distribution of topics in research publications. The method is demonstrated with a pilot study based on approximately a quarter million of publication abstracts in seven diverse research areas. In this example, the method validly reconstructs all obvious topic preferences, for instance, country-dependent language-related preferences. It also produces new insight into country-specific research focuses. It emerges that in all seven subject areas studied, topic preferences are significantly different depending on whether all authors are women, all authors are men, or there are female and male co-authors, with a tendency of male authors towards theoretical core topics, of female authors towards peripheral applied topics, and of mixed-author teams towards modern interdisciplinary topics.
Read full abstract