FluidViews: Drag-and-drop token filters transform multi-view analytics

FluidViews, developed by Bhanu Sunku and presented in a recent arXiv paper (2606.18260), reimagines interactive visual analytics by making filters tangible and directly manipulable. Traditional tools often force analysts to navigate rigid filter panels, causing context switches that break cognitive flow. FluidViews addresses this with two direct-manipulation gestures: Copy-as-Highlight lets users duplicate any visual mark into a highlight token, enabling rapid, transient cross-view comparisons without altering the underlying data. Meanwhile, Drag-as-Filter allows users to pick up a mark and drop it onto a different view to apply context-sensitive filters in place—eliminating menus, modal dialogs, and extra clicks. An optional pop-out micro-view provides on-demand, spatially independent subviews for detailed inspection without disrupting the primary workspace.

The framework is built as a web-based system supporting coordinated multi-view environments across heterogeneous datasets. By embedding lightweight gestures, FluidViews preserves analytic momentum and reduces cognitive overhead during exploratory workflows. The paper demonstrates its application in real-world scenarios, showing how tangible filter objects can transform data exploration into a more fluid, multi-step process. This work contributes to Human-Computer Interaction by challenging the status quo of filter interfaces and offering a new paradigm where filters are not hidden behind menus but are first-class, interactive objects that analysts can see, grab, and reuse. The design is particularly valuable for domains like data journalism, scientific visualization, and business intelligence, where analysts frequently switch between views to uncover insights.