A Design Space of Visualization Tasks

Abstract

Knowledge about visualization tasks plays an important role in choosing or building suitable visual representations to pursue them. Yet, tasks are a multi-faceted concept and it is thus not surprising that the many existing task taxonomies and models all describe different aspects of tasks, depending on what these task descriptions aim to capture. This results in a clear need to bring these different aspects together under the common hood of a general design space of visualization tasks, which we propose in this paper. Our design space consists of five design dimensions that characterize the main aspects of tasks and that have so far been distributed across different task descriptions. We exemplify its concrete use by applying our design space in the domain of climate impact research. To this end, we propose interfaces to our design space for different user roles (developers, authors, and end users) that allow users of different levels of expertise to work with it.

Tasks for Multivariate Network Analysis

“…It is worth reflecting on how the framework presented here corresponds to other recent work on information visualisation tasks. As noted before, Schulz et al describe the visualisation task design space along five dimensions (goal, means, characteristics, target, and cardinality) [10], while Brehmer and Munzner consider three questions (why?, how?, and what?) [11].…”

“…In seminal work, Wehrend and Lewis propose a classification of visualisation methods by considering the entities being studied and tasks performed on the entities [9]. Specifically, they list 11 tasks that are frequently encountered: In other related work on general information visualisation taxonomies, Schulz et al recently proposed a classification of the "design space" of visualisation tasks based on five dimensions (goal; means; characteristics, or level of analysis; target, the parts of the data to be considered; and cardinality, the number of data instances to be considered) [10]. This allows a formal faceted specification of tasks by five-dimensional tuples.…”

A Data Model and Task Space for Data of Interest (DOI) Eye-Tracking Analyses

“…Amar et al proposed more detailed classifications of visualization task types while exploring multidimensional data analysis [7]. More recently, Brehmer et al, Schulz and Nocke, and Rind et al proposed multi-level task frameworks that can aid in creating complete task descriptions regardless of domain [33], [34], [35]. At the same time, specific task taxonomies were defined for visualizations of graph-data [36], group-level graphs [37], multidimensional data visualization [38], [39], and temporal data [40].…”

“…Research on task taxonomies and frameworks converged on the need to describe tasks in terms of users' goals, operands, and objectives [5], [6], [32], [33], [34], [35]. Several frameworks advocate that task descriptions ought to also specify actions or means (i.e., how a task should be achieved) [34], while a few convolute the notions of objectives and actions in a single descriptor [7].…”

“…Several frameworks advocate that task descriptions ought to also specify actions or means (i.e., how a task should be achieved) [34], while a few convolute the notions of objectives and actions in a single descriptor [7]. Since DOI visualizations do not yet exist, and our focus was precisely to motivate the need for DOI analytics and define its requirements, our contribution is limited to defining the space of questions that can be asked of DOI data.…”