Evaluating Multivariate Network Visualization Techniques Using a Validated Design and Crowdsourcing Approach

Abstract

Visualizing multivariate networks is challenging because of the trade-offs necessary for effectively encoding network topology and encoding the attributes associated with nodes and edges. A large number of multivariate network visualization techniques exist, yet there is little empirical guidance on their respective strengths and weaknesses. In this paper, we describe a crowdsourced experiment, comparing node-link diagrams with onnode encoding and adjacency matrices with juxtaposed tables. We find that node-link diagrams are best suited for tasks that require close integration between the network topology and a few attributes. Adjacency matrices perform well for tasks related to clusters and when many attributes need to be considered. We also reflect on our method of using validated designs for empirically evaluating complex, interactive visualizations in a crowdsourced setting. We highlight the importance of training, compensation, and provenance tracking.

reVISit: Looking Under the Hood of Interactive Visualization Studies

“…Studies that evaluate user performance are either geared toward understanding human perceptual and cognitive limits or focused on evaluating an interactive system or technique [25]. Although crowdsourced studies have been widely applied to perceptual experiments and simple visualizations, we recently proposed methods for using crowdsourced empirical studies for sophisticated interactive visualizations [29]. Evaluation of such complex systems, however, comes with new challenges, including the need to train participants on using these visualizations as well as controlling for the effect of different interaction affordances on participant performance.…”

Responsive Matrix Cells: A Focus+Context Approach for Exploring and Editing Multivariate Graphs

“…Formal User Studies For developing our approach, we followed an iterative design process with input from both visualization and HCI experts. Such expert input has proven to be very valuable in similar contexts [54]. Here, we demonstrated by means of a walkthrough how our concepts come into play during an analysis session and provided insights from a preliminary user feedback session.…”

“…As matrices do not explicitly represent the graph nodes, additional means are required to visualize node attributes. Prior research has assessed that a juxtaposed attribute table is a suitable solution [11,54]. An alternative is to calculate a pairwise attributebased similarity measure for nodes and visualize it in one half of the matrix (divided by the diagonal), while the other half still encodes the edges [11].…”

“…A disadvantage of matrices is their quadratic space complexity, which makes visualizing larger graphs demanding [1]. Moreover, matrices are not very well suited for path-related tasks [52,54,79]. A promising approach to mitigate these issues is to combine matrix and node-link representations, as in hierarchical graph maps [1] or Node-Trix [30].…”

Towards Visualizing and Exploring Multivariate Networks on Mobile Devices

“…Nobre et al [12] distinguish between two main categories of layouts, node-link layouts and tabular layouts. In general, node-link diagrams are the more common technique for representing networks, while tabular layouts (e.g., adjacency matrix) may imply a higher cognitive load from the user [13]. Both approaches face scalability issues and can quickly become cluttered, which is particularly relevant in the context of small screen spaces.…”