Visualization systems offer data exploration and are often designed to help domain analysts understand trends, outliers, and patterns in the data. These systems have little to no support for communicating knowledge or insights that are derived through the analysis. Data-driven storytelling, on the other hand, employs specially designed explanatory visualizations or combines a textual narrative alongside a visualization to communicate analysis results to a wider audience, but mostly has no support for exploration. Targeting a sweet spot between explanation and exploration, this doctoral thesis envisions a data representation that integrates explorable visualizations with an automatically generated textual narrative for the dissemination of analysis results to a broad audience.

Grounded in an empirical study on a set of already published data-driven stories, the thesis first describes the interplay of text and visualizations with a focus to derive effective means of achieving a coherent integration between the two media. The thesis then introduces a generic approach for generating an integrated visual and textual representation of data, followed by its instantiations to many diverse application domains and datasets including bivariate geographic data, bibliographic data, knowledge graphs, and source code quality data. Finally, considering the technical challenges and lack of authoring support for the construction of such data representation, the thesis contributes a novel and easy-to-use authoring tool that combines automation with a graphical user interface to establish linking between text and visualizations.

The proposed data representation can still be regarded as a visual analytics solution where text is considered as an active part of the system, just like any other visual element. The text summarizes core findings and hints at notable analysis insights, while intriguing users to explore and even verify the insights. Likewise, while exploring, the text provides additional information on analysis methods and domain-specific terminology, and links back to the explanation.