Visualization for epidemiological modelling: challenges, solutions, reflections and recommendations.

We report on an ongoing collaboration between epidemiological modellers and visualization researchers by documenting and reflecting upon knowledge constructs-a series of ideas, approaches and methods taken from existing visualization research and practice-deployed and developed to support modelling of the COVID-19 pandemic. Structured independent commentary on these efforts is synthesized through iterative reflection to develop: evidence of the effectiveness and value of visualization in this context; open problems upon which the research communities may focus; guidance for future activity of this type and recommendations to safeguard the achievements and promote, advance, secure and prepare for future collaborations of this kind. In describing and comparing a series of related projects that were undertaken in unprecedented conditions, our hope is that this unique report, and its rich interactive supplementary materials, will guide the scientific community in embracing visualization in its observation, analysis and modelling of data as well as in disseminating findings. Equally we hope to encourage the visualization community to engage with impactful science in addressing its emerging data challenges. If we are successful, this showcase of activity may stimulate mutually beneficial engagement between communities with complementary expertise to address problems of significance in epidemiology and beyond. See https://ramp-vis.github.io/RAMPVIS-PhilTransA-Supplement/. This article is part of the theme issue 'Technical challenges of modelling real-life epidemics and examples of overcoming these'.

Towards explainable community finding.

The detection of communities of nodes is an important task in understanding the structure of networks. Multiple approaches have been developed to tackle this problem, many of which are in common usage in real-world applications, such as in public health networks. However, clear insight into the reasoning behind the community labels produced by these algorithms is rarely provided. Drawing inspiration from the machine learning literature, we aim to provide post-hoc explanations for the outputs of these algorithms using interpretable features of the network. In this paper, we propose a model-agnostic methodology that identifies a set of informative features to help explain the output of a community finding algorithm. We apply it to three well-known algorithms, though the methodology is designed to generalise to new approaches. As well as identifying important features for a post-hoc explanation system, we report on the common features found made by the different algorithms and the differences between the approaches. Supplementary Information: The online version contains supplementary material available at 10.1007/s41109-022-00515-6.

Edge-Path Bundling: A Less Ambiguous Edge Bundling Approach.

Edge bundling techniques cluster edges with similar attributes (i.e. similarity in direction and proximity) together to reduce the visual clutter. All edge bundling techniques to date implicitly or explicitly cluster groups of individual edges, or parts of them, together based on these attributes. These clusters can result in ambiguous connections that do not exist in the data. Confluent drawings of networks do not have these ambiguities, but require the layout to be computed as part of the bundling process. We devise a new bundling method, Edge-Path bundling, to simplify edge clutter while greatly reducing ambiguities compared to previous bundling techniques. Edge-Path bundling takes a layout as input and clusters each edge along a weighted, shortest path to limit its deviation from a straight line. Edge-Path bundling does not incur independent edge ambiguities typically seen in all edge bundling methods, and the level of bundling can be tuned through shortest path distances, Euclidean distances, and combinations of the two. Also, directed edge bundling naturally emerges from the model. Through metric evaluations, we demonstrate the advantages of Edge-Path bundling over other techniques.

The Effectiveness of Interactive Visualization Techniques for Time Navigation of Dynamic Graphs on Large Displays.

Dynamic networks can be challenging to analyze visually, especially if they span a large time range during which new nodes and edges can appear and disappear. Although it is straightforward to provide interfaces for visualization that represent multiple states of the network (i.e., multiple timeslices) either simultaneously (e.g., through small multiples) or interactively (e.g., through interactive animation), these interfaces might not support tasks in which disjoint timeslices need to be compared. Since these tasks are key for understanding the dynamic aspects of the network, understanding which interactive visualizations best support these tasks is important. We present the results of a series of laboratory experiments comparing two traditional approaches (small multiples and interactive animation), with a more recent approach based on interactive timeslicing. The tasks were performed on a large display through a touch interface. Participants completed 24 trials of three tasks with all techniques. The results show that interactive timeslicing brings benefit when comparing distant points in time, but less benefits when analyzing contiguous intervals of time.

Event-Based Dynamic Graph Visualisation.

Dynamic graph drawing algorithms take as input a series of timeslices that standard, force-directed algorithms can exploit to compute a layout. However, often dynamic graphs are expressed as a series of events where the nodes and edges have real coordinates along the time dimension that are not confined to discrete timeslices. Current techniques for dynamic graph drawing impose a set of timeslices on this event-based data in order to draw the dynamic graph, but it is unclear how many timeslices should be selected: too many timeslices slows the computation of the layout, while too few timeslices obscures important temporal features, such as causality. To address these limitations, we introduce a novel model for drawing event-based dynamic graphs and the first dynamic graph drawing algorithm, DynNoSlice, that is capable of drawing dynamic graphs in this model. DynNoSlice is an offline, force-directed algorithm that draws event-based, dynamic graphs in the space-time cube (2D+time). We also present a method to extract representative small multiples from the space-time cube. To demonstrate the advantages of our approach, DynNoSlice is compared with state-of-the-art timeslicing methods using a metrics-based experiment. Finally, we present case studies of event-based dynamic data visualised with the new model and algorithm.

GrouseFlocks: steerable exploration of graph hierarchy space.

Several previous systems allow users to interactively explore a large input graph through cuts of a superimposed hierarchy. This hierarchy is often created using clustering algorithms or topological features present in the graph. However, many graphs have domain-specific attributes associated with the nodes and edges, which could be used to create many possible hierarchies providing unique views of the input graph. GrouseFlocks is a system for the exploration of this graph hierarchy space. By allowing users to see several different possible hierarchies on the same graph, the system helps users investigate graph hierarchy space instead of a single fixed hierarchy. GrouseFlocks provides a simple set of operations so that users can create and modify their graph hierarchies based on selections. These selections can be made manually or based on patterns in the attribute data provided with the graph. It provides feedback to the user within seconds, allowing interactive exploration of this space.

A Vector Field Design Approach to Animated Transitions.

Animated transitions can be effective in explaining and exploring a small number of visualizations where there are drastic changes in the scene over a short interval of time. This is especially true if data elements cannot be visually distinguished by other means. Current research in animated transitions has mainly focused on linear transitions (all elements follow straight line paths) or enhancing coordinated motion through bundling of linear trajectories. In this paper, we introduce animated transition design, a technique to build smooth, non-linear transitions for clustered data with either minimal or no user involvement. The technique is flexible and simple to implement, and has the additional advantage that it explicitly enhances coordinated motion and can avoid crowding, which are both important factors to support object tracking in a scene. We investigate its usability, provide preliminary evidence for the effectiveness of this technique through metric evaluations and user study and discuss limitations and future directions.

TopoLayout: multilevel graph layout by topological features.

We describe TopoLayout, a feature-based, multilevel algorithm that draws undirected graphs based on the topological features they contain. Topological features are detected recursively inside the graph, and their subgraphs are collapsed into single nodes, forming a graph hierarchy. Each feature is drawn with an algorithm tuned for its topology. As would be expected from a feature-based approach, the runtime and visual quality of TopoLayout depends on the number and types of topological features present in the graph. We show experimental results comparing speed and visual quality for TopoLayout against four other multilevel algorithms on a variety of data sets with a range of connectivities and sizes. TopoLayout frequently improves the results in terms of speed and visual quality on these data sets.

Evaluation of Graph Sampling: A Visualization Perspective.

Graph sampling is frequently used to address scalability issues when analyzing large graphs. Many algorithms have been proposed to sample graphs, and the performance of these algorithms has been quantified through metrics based on graph structural properties preserved by the sampling: degree distribution, clustering coefficient, and others. However, a perspective that is missing is the impact of these sampling strategies on the resultant visualizations. In this paper, we present the results of three user studies that investigate how sampling strategies influence node-link visualizations of graphs. In particular, five sampling strategies widely used in the graph mining literature are tested to determine how well they preserve visual features in node-link diagrams. Our results show that depending on the sampling strategy used different visual features are preserved. These results provide a complimentary view to metric evaluations conducted in the graph mining literature and provide an impetus to conduct future visualization studies.

Smashing peacocks further: drawing quasi-trees from biconnected components.

Quasi-trees, namely graphs with tree-like structure, appear in many application domains, including bioinformatics and computer networks. Our new SPF approach exploits the structure of these graphs with a two-level approach to drawing, where the graph is decomposed into a tree of biconnected components. The low-level biconnected components are drawn with a force-directed approach that uses a spanning tree skeleton as a starting point for the layout. The higher-level structure of the graph is a true tree with meta-nodes of variable size that contain each biconnected component. That tree is drawn with a new area-aware variant of a tree drawing algorithm that handles high-degree nodes gracefully, at the cost of allowing edge-node overlaps. SPF performs an order of magnitude faster than the best previous approaches, while producing drawings of commensurate or improved quality.

A Simple Approach for Boundary Improvement of Euler Diagrams.

General methods for drawing Euler diagrams tend to generate irregular polygons. Yet, empirical evidence indicates that smoother contours make these diagrams easier to read. In this paper, we present a simple method to smooth the boundaries of any Euler diagram drawing. When refining the diagram, the method must ensure that set elements remain inside their appropriate boundaries and that no region is removed or created in the diagram. Our approach uses a force system that improves the diagram while at the same time ensuring its topological structure does not change. We demonstrate the effectiveness of the approach through case studies and quantitative evaluations.

AmbiguityVis: Visualization of Ambiguity in Graph Layouts.

Node-link diagrams provide an intuitive way to explore networks and have inspired a large number of automated graph layout strategies that optimize aesthetic criteria. However, any particular drawing approach cannot fully satisfy all these criteria simultaneously, producing drawings with visual ambiguities that can impede the understanding of network structure. To bring attention to these potentially problematic areas present in the drawing, this paper presents a technique that highlights common types of visual ambiguities: ambiguous spatial relationships between nodes and edges, visual overlap between community structures, and ambiguity in edge bundling and metanodes. Metrics, including newly proposed metrics for abnormal edge lengths, visual overlap in community structures and node/edge aggregation, are proposed to quantify areas of ambiguity in the drawing. These metrics and others are then displayed using a heatmap-based visualization that provides visual feedback to developers of graph drawing and visualization approaches, allowing them to quickly identify misleading areas. The novel metrics and the heatmap-based visualization allow a user to explore ambiguities in graph layouts from multiple perspectives in order to make reasonable graph layout choices. The effectiveness of the technique is demonstrated through case studies and expert reviews.

Tugging graphs faster: efficiently modifying path-preserving hierarchies for browsing paths.

Many graph visualization systems use graph hierarchies to organize a large input graph into logical components. These approaches detect features globally in the data and place these features inside levels of a hierarchy. However, this feature detection is a global process and does not consider nodes of the graph near a feature of interest. TugGraph is a system for exploring paths and proximity around nodes and subgraphs in a graph. The approach modifies a pre-existing hierarchy in order to see how a node or subgraph of interest extends out into the larger graph. It is guaranteed to create path-preserving hierarchies, so that the abstraction shown is meaningful with respect to the underlying structure of the graph. The system works well on graphs of hundreds of thousands of nodes and millions of edges. TugGraph is able to present views of this proximal information in the context of the entire graph in seconds, and does not require a layout of the full graph as input.

A screening test for the determination of ethylene sensitivity.

Ethephon, which releases ethylene within plant tissues after application, was chosen to perform assessments of the relative sensitivity of crops to ethylene and to determine which stages of plant development were most sensitive. The species chosen were: barley, wheat, oats, canola and field pea, all of which are important crops in the province of Alberta, Canada. Plants were treated with ethephon at one of 7 different stages. Plants were assessed for their vegetative and reproductive growth, including height, biomass, yield and seed quality. Visual symptoms were photographed and documented to compare them with symptoms caused by ethylene applied as a gas. It was concluded that in barley, wheat and canola the late vegetative and early reproductive stages were most sensitive, at least when sensitivity was defined as reductions in yield and quality. As for field pea, ethephon had no effect on yield but did cause increased numbers of pods, which in certain conditions could lead to increased yields. Significant effects on vegetative growth were only observed in the early vegetative stages of development but with no effects on yield. The screening protocol successfully identified sensitive cultivars and growth stages for further investigation of the effects of ethylene exposure.