Data from social media platforms and online communitieshave fueled the growth of computational social science. In this chap-ter, we use computational analysis to characterize the state of researchon social media and demonstrate the utility of such methods. First,we discuss how to obtain datasets from the APIs published by manysocial media platforms. Then, we perform some of the most widelyused computational analyses on a dataset of social media scholarship weextract from the Scopus bibliographic database’s API. We apply threemethods: network analysis, topic modeling using latent Dirichlet allo-cation, and statistical prediction using machine learning. For each tech-nique, we explain the method and demonstrate how it can be used todraw insights from our dataset. Our analyses reveal overlapping schol-arly communities studying social media. We find that early social me-dia research applied social network analysis and quantitative methods,but the most cited and influential work has come from marketing andmedical research. We also find that publication venue and, to a lesserdegree, textual features of papers explain the largest variation in incom-ing citations. We conclude with some consideration of the limitationsof computational research and future directions.
Foote, J., Shaw, A., & Hill, B. M. (2018). A computational analysis of social media scholarship. The SAGE handbook of social media, 111-134.http://mako.cc/academic/foote_shaw_hill-computational_analysis_of_social_media.pdf