Uncertainty-Aware Data Transformations for Collaborative Reasoning

The goal of this research is to develop mathematical foundations of uncertainty-aware data transformation methods to facilitate trustworthy collaborative reasoning using visual meaning. Due to data collected for analysis is often incomplete, contradictory or riddled with errors, the insight gained by its visualization becomes inevitably uncertain. In most cases, however, this uncertainty information is not available to the user, who must grossly estimate or simply ignore, the confidence levels of the resulting visualization. I will present an uncertainty-aware
framework for evaluating the visual analytics process. Via uncertainty propagation and aggregation, we can then measure and quantify the uncertainty on raw and derived data throughout the process. With such a framework, analysts will have the ability to observe the confidence levels on the
gained insight, identify transformations and sources with higher uncertainty, and assess “what-if ” scenarios.