Cognitive Stages in Visual Data Exploration

Abstract

Data exploration requires forming analysis goals, planning actions and evaluating results effectively, all of which are complex cognitive activities. Therefore, the data exploration and analysis process can be improved through a principled and comprehensive approach to analyzing the cognitive activities of the user given a data exploration tool. However, many taxonomies and evaluations focus on a specific tool or specific design guides instead of cognitive activities comprehensively. In this paper, we first present the Cognitive Exploration Framework that identifies six stages of cognitive activities in visual data exploration. These stages are a combination of two activities---planning and assessing---across data analysis, interaction, and visualization. Cognitive barriers in each stage can lower the success and speed of data exploration. The framework also identifies the factors of decision-making, existing knowledge and motivation that influence cognitive activities. We argue that cognitive stages can be supported by improving the design of tools rather than their computing capabilities. We demonstrate how the framework clarifies the structured relationship between design guides to specific cognitive stages. In particular, the framework can also be used to guide evaluation of data exploration tools. To reveal cognitive barriers in each stage, we focused on the failures instead of success stories, and on motivating self-driven open-ended exploration instead of using benchmarked tasks on fixed datasets. With these goals, we studied short-term casual use of an exploratory tool by novices with limited training. Our results reveal cognitive barriers across all stages. We also discuss directions for future research and applications.