Big data techniques applied to the study and characterisation of scientific activity on social media

Resumen

The advent of social media has spawned an entire digital ecosystem for communication and information management. This change has had a profound effect on science and the way its results are published and disseminated. Twitter, Wikipedia, and news outlets are now the visible heads of an extensive number of channels for scientific communication, integrating and making the discourse and dissemination of scientific results visible to society as a whole. This has led to the exploration of how science is consumed in such environments and the attention it captures beyond the realm of academia. However, a lack of depth and exploitation of the media studied beyond counting mentions of scholarly outputs has been identified, along with putting the activity around science into greater context. There also exists the unexplored platforms and limited adaptation of traditional methods of scientometrics for the quantitative study of science. This thesis aims to address these challenges to delve into the potential of massive social media data and the heterogeneity of social media for the study of science by combining data science and scientometrics. As a result, proposals for conceptual and methodological frameworks for the use and mapping of social media data have been developed. For this purpose, classic scientometric techniques have been adapted for social network analysis, and new methods have been proposed for the creation of scientific maps that combine social and semantic information. This allows the identification of knowledge structures established through social activity and the identification of cognitive communities of social actors. Furthermore, the methodological proposals have been put into practice through case studies and large-scale studies to validate them and provide novel results on the discussion and dissemination of science on Twitter and Wikipedia, particularly in comparison to academia.