Sensemaking in Visual Analytics: Processes and Challenges

Abstract

Since Visual Analytic systems support human sensemaking it is essential that such systems are designed with characteristics of this process in mind. Drawing on our previous work with lawyers and reports from experienced fraud investigators we describe the nature of the cognitive work to be supported. We describe the cognitive work domain in terms of its data characteristics, and develop a model of the sensemaking as basis for discussing a distinction between 'naturalistic' and 'normative' sensemaking with a particular emphasis on inference types and the potential for bias. We also report results from a questionnaire-based case study designed to elicit memorable incidents from fraud investigators' experiences. Given the legal context the case study exemplifies skills and strategies that are necessary in order to achieve normative and defensible sensemaking under pressure of highvolume datasets.