Joost Gadellaa
Utrecht University
Capita Selecta project
Business Informatics
Privacy is a versatile field of research, studied in a multitude of disciplines. A novel way to explore how research concepts like these are embedded in scientific literature is through the use of topic models. This study aims to investigate the usefulness of such modelling techniques. We use a latent Dirichlet allocation (LDA) topic model to analyse the title, abstract and keywords of 83,159 works in the research field of privacy, evaluating the model both on its own as well as in comparison to a previously made citation network analysis. We show that, although the resulting topic model can be meaningfully labeled, interpretation yields little insights. When comparing the model to the citation network analysis, remarkable similarities between the resulting classifications of these very different methods show. However, many methodological caveats pose a threat to conclusion validity, and computational costs limit possibilities for additional exploration. This method needs further research in order to generate insights, instead of just being an accurate representation of the data.
Keywords : privacy, topic model, LDA, network analysis, bibliometrics