III:Small:Communicating and Resolving Ambiguity in Collaborative Visual Analytics

Collaborative work is essential for solving complex problems and making critical decisions. This is evident in domains like national security, where intelligence agencies work together; scientific research, where different fields combine their knowledge; and public health, where coordinated strategies address health crises. Effective collaboration relies on clear communication, but this remains challenging due to the inherently complex and expressive nature of human interaction.

Even with advances in cloud storage, telepresence, and virtual workspaces, achieving complete clarity is difficult. The integration of data and artificial intelligence further complicates matters, introducing issues such as uncertainty, trust deficits, and varying interpretations of actions. Ambiguity, where multiple interpretations of the same data exist, can significantly impede a team's ability to achieve a shared understanding and make informed decisions.

Since collaboration often relies on visual mediums like displays, video conferences, and documents, there is substantial potential for visually conveying the nature and degree of ambiguity inherent in collaborative tasks. This project aims to develop advanced techniques and tools to improve the management and resolution of ambiguity within collaborative visual analytics, enhancing visual communication to ensure that all team members can comprehend and agree on the information presented.

By addressing these challenges, the project seeks to improve collaborative efforts across various domains, leading to better-informed decisions and substantial societal benefits.

This project aims to provide teams with methods to more effectively identify, visualize, and track ambiguity in collaborative analytics. The technical approach is divided into three main aspects. First, the project will develop techniques to identify the sources of ambiguity in collaborative analytics tasks, including data, AI/ML models, visualizations, and analytical narratives.

This will involve pinpointing instances where data and results lead to multiple interpretations, influencing analytical inferences and decisions. Second, the project will design new domain-relevant visual metaphors to represent ambiguity for effective communication. These visual metaphors will be specifically tailored to convey ambiguity within domains involving critical decision-making, and the project aims to develop a general set of design guidelines for visually representing ambiguity.

Metaphors should highlight multiple interpretations from different sources, but it is crucial to balance metaphorical and abstract data representation to ground the interpretation without overwhelming the abstract representation. Third, the project will develop techniques empowered by large language models to track, interpret, and visually articulate the downstream effects of ambiguity on collaborative decision-making processes.

By referring to domain-specific knowledge bases (e.g., ontologies), the project will interpret verbal markers in natural language communication within the target domain and visually expose potential ambiguities in the interpretation. The project will employ graph representations and explore graph-based methods to capture interactions between the analyst and the analyzed information, the operations performed, and the visualizations used.

This simplifies conflict resolution between multiple interpretations and makes it more computationally feasible. These three research aims will be complemented by a comprehensive evaluation plan involving both formative and summative studies, as well as longitudinal studies with domain experts from emergency management and clinical decision-making. The focused research efforts will introduce a new paradigm in the visualization field to convey and manage ambiguity.

Such advancement will enhance the clarity, efficiency, and effectiveness of team efforts and decision-making in an increasingly complex, AI-powered world.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.