VizCouple: Automating Visualization Generation and Integration from Data-Rich Documents

Data-rich documents are increasingly important for information presentation, as data is essential to societal decision-making and everyday communication. However, plain text is often insufficient for expressing complex data relationships, while traditional visualization tools, though powerful, are difficult to scale, requiring expertise and costly implementation. Even when charts are incorporated, separation from surrounding narrative can disrupt reading flow, making it harder to stay focused and understand key points. To address these issues, we present VizCouple, a system that automatically generates interactive charts and links them closely with relevant text. VizCouple embeds character-sized charts into the narrative, enabling smooth two-way navigation between text and visuals. In a user study, versus a plain-text interface, VizCouple reduced average task completion time for data identification and sorting by 30% and improved accuracy on complex analytical tasks by 14.29%. Participants reported lower cognitive effort and welcomed its embedded visual design and interactive links.