Effective Visual Analytics for Exploring Argumentation and Deliberation in Text Corpora

Argumentation is a natural part of human communication and a focus of interest for social scientists and linguists. Nevertheless, few visualization and interaction techniques visually support social scientists working with argumentative and deliberative texts. To address this research gap, this thesis investigates new techniques for visually comparing and analyzing the argumentative and deliberative texts of a whole corpus in order to facilitate scholarly work and enable exploratory research on such text corpora. The investigation proceeds from three perspectives: Visual techniques to express both the hierarchical relationships of multiple argumentation structures and the order of their arguments, supporting detailed interactive analysis; Visual approaches for analyzing deliberative processes in multiple hierarchical thread structures of online discussions; Interaction techniques to smoothly transition between scatter plots and parallel coordinates, enabling the exploration of statistical measures of argumentation structures in relation to other factors such as topics and human values. The work advances visualization research through several contributions: novel aggregation and visualization techniques for tree sets; a scalable rose-chart-like glyph design for representing node-related attributes in aggregated tree sets, along with a related perception study; and a visual computational notebook for interactive, data-driven research. Furthermore, a seamless geometric transition between scatter plots and parallel coordinates facilitates complex analyses, while additional point representatives in parallel coordinates enhance cluster and correlation identification. These contributions not only advance visualization research but also provide the social sciences with new visual analysis tools that implement the developed techniques to evaluate their practical value. Many of these techniques enable new ways of conducting research -- exploratory as well as hypothesis-driven -- offering fresh insights from existing data and a broader range of analysis possibilities for future study results. In teaching, these results help visually explain argumentation structures, deliberative processes, and the relationship between parallel coordinates and scatter plots.