Active Visual Analytics: Assisted Data Discovery in Interactive Visualizations via Active Search

We propose the Active Visual Analytics technique (ActiveVA), an augmentation of interactive visualizations with active search to aid data foraging and information extraction. We accomplish this by integrating an active search algorithm into the visual interface, which leads users to explore promising data points on a visualization and updates suggestions upon observing user feedback. Using a fictitious epidemic dataset published by the VAST community, we conduct two evaluations. First, we present simulation results that demonstrate the effectiveness of active search in data discovery. Second, we show how a human-computer partnership based on ActiveVA can result in more meaningful interactions during interactive visual exploration and discovery with a crowd-sourced user study. Finally, we outline open research questions regarding human factors in active search and the implications of this line of research in real-world scenarios such as visual drug discovery and intelligence analysis.