Uncertainty Propagation and Trust Building in Visual Analytics

Abstract

Visual analytics combines human and machine abilities to generate new knowledge from data. Within this process, uncertainty often plays an important role in hindering the sensemaking process and analysis tasks. On the machine side, uncertainty builds up from the data source level to the visual output. On the human side, these uncertainties often result in “lack of knowledge or trust” or “overtrust.” Such human’s biased interpretation can be resolved if we can measure uncertainties and users’ trust at each stage and provide proper mitigation in time. We believe that we can achieve this by tracing data provenance and analytic provenance accurately and reflecting them on the system output. Therefore, our first goal is to identify the roles of uncertainty and trust along the entire visual analytics knowledge generation process. In addition, we aim to capture how uncertainty and trust can be derived from data and analytic provenance. In this workshop, we introduce a framework that describes the roles of uncertainty and trust, and introduce open research questions with potential solutions.