Activity-Centered Domain Characterization for Problem-Driven Scientific Visualization

Abstract

Although visualization design models exist in the literature in the form of higher-level methodological frameworks, these models do not present a clear methodological prescription for the domain characterization step. This work presents a framework and end-to-end model for requirements engineering in problem-driven visualization application design. The framework and model are based on the activity-centered design paradigm, which is an enhancement of human-centered design. The proposed activity-centered approach focuses on user tasks and activities, and allows an explicit link between the requirements engineering process with the abstraction stage—and its evaluation—of existing, higher-level visualization design models. In a departure from existing visualization design models, the resulting model: assigns value to a visualization based on user activities; ranks user tasks before the user data; partitions requirements in activity-related capabilities and nonfunctional characteristics and constraints; and explicitly incorporates the user workflows into the requirements process. A further merit of this model is its explicit integration of functional specifications, a concept this work adapts from the software engineering literature, into the visualization design nested model. A quantitative evaluation using two sets of interdisciplinary projects supports the merits of the activity-centered model. The result is a practical roadmap to the domain characterization step of visualization design for problem-driven data visualization. Following this domain characterization model can help remove a number of pitfalls that have been identified multiple times in the visualization design literature.

A Review and Characterization of Progressive Visual Analytics

“…This characterization mirrors recent research in activity-centered visualization design that assesses a visualization based on the tasks and activities it supports [20], instead of just looking at how faithful it represents the data (cp. representational primacy [21]).…”

Visual Analysis of a Smart City’s Energy Consumption

“…Our design process followed an Activity-Centered-Design paradigm for visualization [16], which is an extension of the classic Human Centered Design paradigm in visualization design. The approach places particular emphasis on functional specifications and on user workflows.…”

“…The first stage of design, requirement engineering, started with several face-to-face semi-structured interviews with two energy researchers. Because Activity-Centered-Design [16] focuses on activities, not the individual person, no personal data was collected from the energy researchers. The interviews established: who the potential users of the visualization would be (energy researchers and policymakers; with the clear objective or reaching the broader population); a prioritized list of the main analysis tasks and workflows; the data sources and flow of data through the process; and non-functional requirements such as web-access and support for large displays.…”

“…We analyzed the requirements resulting from the interviews along the Activity-Centered-Design components of tasks, usage, data, flow, and nonfunctional requirements [16]. The data requirements are described in detail in the following section.…”

“…The data requirements are described in detail in the following section. We wrote the resulting functional specifications as scenarios [16]. A first set of scenarios was centered around policymakers and energy researcher characters.…”