Multivariate graphs are an important data type in many high-impact application areas. Yet we know little about the underlying principles of how to effectively visualize them.  This project established a set of validated, foundational principles for visualizing multivariate graphs using a structured, methodological research approach. We describe the core contributions of our work in the following paragraphs.

Multivariate graphs that consist of nodes that are spatial constrained -- such nodes with a geospatial location -- are difficult to visualize due to the inability to move nodes to reduce visual clutter. In the Poemage software we tackled this challenge by developing a novel technique that supports visualization of paths through spatially constrained graphs. 

A key task in multivariate graph analysis is the exploration of connectivity, to, for example, analyze how signals flow through neurons, or to explore how well different cities are connected by flights. In the Graffinity software we introduced a technique called the connectivity matrix that provides an overview of the connectivity and reveal details on demand.

Multivariate networks are a natural way of thinking about many datasets. The data on which a network is based, however, is rarely collected in a form that suits the analysis process, making it necessary to create and reshape networks. Data wrangling is widely acknowledged to be a critical part of the data analysis pipeline, yet interactive network wrangling has received little attention in the visualization research community. In the Origraph software system we implement a set of operations that are important for wrangling multivariate network datasets. 

Visualization of multivariate networks is challenging, especially when both the topology of the network and the attributes need to be considered concurrently. Through an analysis of all existing visualization techniques for multivariate networks we developed a characterizing design space of possible techniques. We analyzed current practices and classify techniques along four axes: layouts, view operations, layout operations, and data operations. We also provide an analysis of tasks specific to multivariate networks and give recommendations for which technique to use in which scenario.

Through our technical research we additionally developed several methodological techniques to improve design-oriented visualization research practices.

Applied visualization researchers often work closely with domain collaborators to explore new and useful applications of visualization. The early stages of collaborations are typically time consuming for all stakeholders as researchers piece together an understanding of domain challenges from disparate discussions and meetings. A number of recent projects, however, report on the use of creative visualization-opportunities (CVO) workshops to accelerate the early stages of applied work, eliciting a wealth of requirements in a few days of focused work. Yet, there is no established guidance for how to use such workshops effectively. Through a two-year collaboration in which we analyzed the use of 17 workshops in 10 visualization contexts we developed a framework for CVO workshops that 1) identifies a process model for using workshops; 2) describes a structure of what happens within effective workshops; 3) recommends 25 actionable guidelines for future workshops; and 4) presents an example workshop and workshop methods.

In applied visualization research, artifacts are shaped by a series of small design decisions, many of which are evaluated quickly and informally via methods that often go unreported and unverified. Such design decisions are influenced not only by visualization theory, but also by the people and context of the research. While existing applied visualization models support a level of reliability throughout the design process, they fail to explicitly address the influence of the research context in shaping the resulting design artifacts. In this work, we look to action design research (ADR) for insight into filling this gap. In particular, ADR offers a framework along with a set of guiding principles for navigating and capitalizing on the disruptive, subjective, human-centered nature of applied design research, while aiming to ensure reliability of the process and design.

The nested blocks and guidelines model for describing visualization design decisions extends previous models to provide explicit mechanisms that capture and discuss design decision rationale. Blocks are the outcomes of design decisions throughout the design process, and guidelines discuss relationships between these blocks. Using the NBGM we are able to more concretely identify possible weaknesses in exisitng and new guidelines, clarify assumptions that require further evaluation, and provide feedback on the rigor and validity of visualization research results.

Last Modified: 10/15/2020
Modified by: Miriah Meyer