Fig 8 - uploaded by Yong Wang
Content may be subject to copyright.
![Detailed comparison of three graph layouts for the same data [19]. The rows from top to bottom represent views of the original layout, edge crossing (EC), convex hull (CH), edge length ambiguity (EL), node-edge occlusion (NEO) and autocorrelation-based ambiguity (ACA) metrics.](profile/Yong-Wang-149/publication/281467141/figure/fig5/AS:614017152323596@1523404505997/Detailed-comparison-of-three-graph-layouts-for-the-same-data-19-The-rows-from-top-to.png)
Detailed comparison of three graph layouts for the same data [19]. The rows from top to bottom represent views of the original layout, edge crossing (EC), convex hull (CH), edge length ambiguity (EL), node-edge occlusion (NEO) and autocorrelation-based ambiguity (ACA) metrics.
Source publication
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 netwo...
Contexts in source publication
Context 1
... net- work data [19] representing the network of American football games between Division IA colleges during the regular season in 2000. The network contains 115 nodes and 613 edges. We use Gephi to generate three force-directed graph layouts: ForceAtlas2 with LinLog mode en- abled (FA2+LinLog) [29], FR [14], and Hu's method (HU) [26] shown in Fig. 8. The same parameter settings are used for all three heatmaps. We can see that the layout of FA2+LinLog has advantages in preserv- ing the community structures indicated by the autocorrelation-based metric and convex hull, as shown in the "ACA" row and the "CH" row of Fig. 8. This is not surprising, considering that LinLog is good at ...
Context 2
... en- abled (FA2+LinLog) [29], FR [14], and Hu's method (HU) [26] shown in Fig. 8. The same parameter settings are used for all three heatmaps. We can see that the layout of FA2+LinLog has advantages in preserv- ing the community structures indicated by the autocorrelation-based metric and convex hull, as shown in the "ACA" row and the "CH" row of Fig. 8. This is not surprising, considering that LinLog is good at preserving cluster structure in graphs [29] and this property can also be easily seen in the original layout. The visual overlap ambiguity views based on autocorrelation further inform users that FR incurs more vi- sual overlaps between communities than the other ...
Context 3
... metric clearly shows that several nodes are near the regions of other commu- nities. This may affect user's perception of these communities, but the problem cannot be easily identified in the original graph layout. The node occlusion views show that all three layouts have little to no node occlusion and thus are not presented in Fig. 8. When examin- ing edge crossings, FA2+LinLog has relatively less ambiguity overall when compared with FR and HU. This finding is supported by both the global metric value and the heatmap. However, the crossings can be serious in local regions, impeding a users understanding the connec- tivity between nodes. When considering the edge ...
Similar publications
The visual complexity of a graph drawing is defined as the number of geometric objects needed to represent all its edges. In particular, one object may represent multiple edges, e.g., one needs only one line segment to draw two collinear incident edges. We study the question if drawings with few segments have a better aesthetic appeal and help the...
The ability to represent relational data in the form of drawings or graphs is a powerful tool which allows us to perform analysis through visual exploration. Several data presentation problems require the drawing or display of graphs; examples include circuit schematics and software engineering diagrams. Force-directed graph drawing algorithms are...
Citations
... Similarly, to highlight common visual ambiguities such as visual overlap between community structures, an issue caused by the problematic hairball layout described in subsection 2.1.1, Wang et al. [Wan+15] encoded known clustering information in the visualisations to highlight visual cluster boundaries. Previous work also presented evaluations of task-performance for such group-level visualisations [Sak+14; OJK18; VBW17]. ...
... This network has been previously evaluated in social network visualisation literature. For example, Wang et al. [Wan+15] proposed methods to highlight visual ambiguities as a result of highly connected network structure that leads to visual cluster overlap. They encoded cluster information in the layouts based on pre-computed community detection to better show cluster boundaries. ...
Visualisation has become increasingly important in helping people understand and gain insights from data; and, ultimately, to make better decisions. Traditional visualisations are designed to be shown on a flat surface (screen or page) but most data is not “flat”. For example, the surface of the earth exists on a sphere, however, when that surface is presented on a flat map, key information is hidden, such as geographic paths on the spherical surface being wrapped across the boundaries of the flat map. Similarly, cyclical time-series data has no beginning or end. When such cyclical data is presented on a traditional linear chart, the viewer needs to perceive continuity of the visualisation across the chart’s boundaries. Mentally reconnecting the chart across such a boundary may induce additional cognitive load. More complex data such as a network diagram with hundreds or thousands of links between data points leads to a densely connected structure that is even less “flat” and needs to wrap around in multiple dimensions.
To improve the usability of these visualisations, this thesis explores a novel class of interactive wrapped data visualisations, i.e., visualisations that wrap around continuously when interactively panned on a two-dimensional (2D) projection of surfaces of 3D shapes, specifically, cylinder, torus, or sphere. We start with a systematic exploration of the design space of interactive wrapped visualisations, characterising the visualisations that help people understand the relationship within the data and benefit from being understood as ‘cylindrical’, ‘toroidal’, or ‘spherical’. Subsequently, we investigate the design, development and implementation of a series of wrappable visualisations for cyclical time series, network, and geographic data. We show that these interactive visualisations better preserve the spatial relations in the case of geospatial data, and better reveal the data’s underlying structure in the case of abstract data such as networks and cyclical time series. Furthermore, to assist future research and development, we contribute layout algorithms and toolkits to help create pannable wrapped visualisations.
... Similarly, to highlight common visual ambiguities such as visual overlap between community structures, an issue caused by the problematic hairball layout described in subsection 2.1.1, Wang et al. [Wan+15] encoded known clustering information in the visualisations to highlight visual cluster boundaries. Previous work also presented evaluations of task-performance for such group-level visualisations [Sak+14; OJK18; VBW17]. ...
... This network has been previously evaluated in social network visualisation literature. For example, Wang et al. [Wan+15] proposed methods to highlight visual ambiguities as a result of highly connected network structure that leads to visual cluster overlap. They encoded cluster information in the layouts based on pre-computed community detection to better show cluster boundaries. ...
Traditional visualisations are designed to be shown on a flat surface (screen or page) but most data is not "flat". For example, the surface of the earth exists on a sphere, however, when that surface is presented on a flat map, key information is hidden, such as geographic paths on the spherical surface being wrapped across the boundaries of the flat map. Similarly, cyclical time-series data has no beginning or end. When such cyclical data is presented on a traditional linear chart, the viewer needs to perceive continuity of the visualisation across the chart's boundaries. Mentally reconnecting the chart across such a boundary may induce additional cognitive load. More complex data such as a network diagram with hundreds or thousands of links between data points leads to a densely connected structure that is even less "flat" and needs to wrap around in multiple dimensions. To improve the usability of these visualisations, this thesis explores a novel class of interactive wrapped data visualisations, i.e., visualisations that wrap around continuously when interactively panned on a two-dimensional projection of surfaces of 3D shapes, specifically, cylinder, torus, or sphere. We start with a systematic exploration of the design space of interactive wrapped visualisations, characterising the visualisations that help people understand the relationship within the data. Subsequently, we investigate a series of wrappable visualisations for cyclical time series, network, and geographic data. We show that these interactive visualisations better preserve the spatial relations in the case of geospatial data, and better reveal the data's underlying structure in the case of abstract data such as networks and cyclical time series. Furthermore, to assist future research and development, we contribute layout algorithms and toolkits to help create pannable wrapped visualisations.
... This additional visualization reduces the visualization uncertainty that is inherent to the dimensionality reduction underlying the Time Curves. Works addressing visualization of uncertainty arising from dimensionality reduction [24,43] explicitly show the error of these methods. ...
This paper provides a brief overview of uncertainty visualization along with some fundamental considerations on uncertainty propagation and modeling. Starting from the visualization pipeline, we discuss how the different stages along this pipeline can be affected by uncertainty and how they can deal with this and propagate uncertainty information to subsequent processing steps. We illustrate recent advances in the field with a number of examples from a wide range of applications: uncertainty visualization of hierarchical data, multivariate time series, stochastic partial differential equations, and data from linguistic annotation.
... We adopt the ten metrics in the evaluation framework proposed by Haleem et al. [29] and two community-related metrics proposed by Wang et al. [77]. These metrics are also widely used to evaluate other graph layout algorithms. ...
Node-link diagrams are widely used to visualize graphs. Most graph layout algorithms only consider graph topology or node attributes for aesthetic goals (e.g., with fewer edge crossings and node occlusions), resulting in information loss and waste. Existing hybrid approaches that bind the two perspectives mostly build layouts on top of the attribute-based communities to better satisfy exploration goals. However, they usually suffer from high human dependency, input restriction, and loosely-coupled bindings of topology and attributes, thus may have limited substantial improvements to the layout quality. In this paper, we propose an embedding-based graph exploration pipeline to enjoy the best of both graph topology and node attributes. First, we leverage embedding algorithms for attributed graphs to encode the two perspectives into latent space. Then, we present an embedding-driven graph layout algorithm, GEGraph, which can achieve aesthetic layouts with better community preservation to support an easy interpretation of the graph structure. Next, graph explorations are extended based on the generated graph layout and insights extracted from the embedding vectors. Illustrated with examples, we build a layout-preserving aggregation method with Focus+Context interaction and a related nodes searching approach with multiple proximity strategies. Finally, we conduct quantitative and qualitative evaluations, a user study, and two case studies to validate our approach.
... Some previous works [41], [42] used edge bundling to simplify a graph. Wang et al. [43] proposed a visual technique to reveal ambiguities. Langevin [44] presented a visual analytical approach to leverage cluster computing by retrieving hierarchical communities from the input data. ...
Bitcoin is gaining ever increasing popularity. However, professional skills are required if people want to check bitcoin transaction information from the blockchain. As pointed out in a recent study [1], there is a lack of tools to support effective interactive investigation of bitcoin transactions. Therefore, we present a novel visualization system, BitAnalysis , for interactive bitcoin wallet investigation. The analytical and visualization functions of BitAnalysis are defined and developed by following the advice and requirements of a group of entrepreneurs and regulators of bitcoin-related business. BitAnalysis provides a rich set of functions and intuitive visual interfaces for the users, such as law-enforcement officers and regulators, to effectively visualize and analyze the transactions of a bitcoin wallet (i.e., a cluster of bitcoin addresses) and its related wallets, to track the flow of bitcoins, and to identify wallet correlation using our novel clustering functions. To achieve these functions, we have designed new visualization techniques for presenting bitcoin transactions information and introduced the connection diagram and bitcoin flow map as new ways of analyzing, tracking and monitoring the trading activities of a cluster of closely related wallets. We also present an extensive user study that validated the effectiveness and usability of BitAnalysis .
... Autocorrelation metric in [32] with MCL clustering [29]. Neighborhood preservation* As defined in [17]. ...
... -Layout conventions The iStar model is a type of graphlike model, so we also refer to layout conventions that are generally applied to graph visualization. In this paper, we take three aesthetic metrics into account and generate three layout requirements correspondingly: avoidance of overlapping elements (GR2), crossing edges (GR4), and overlapping communities (SDR2), since they are widely used aesthetic criteria when dealing with graph layouts [11,27,37]. Moreover, to better present the layout results, we also consider the center-aligned fashion (GR1) as a convention. ...
Bridging the gap between academia and industry is an important issue to promote the practicality of i* framework. Researchers have been dealing with this issue from various perspectives, such as simplifying the meta-models or modeling processes of i* framework. In this paper, we exclusively focus on the scalability issue in laying out large-scale i* models and propose a two-level layout approach to automatically lay out i* models in an efficient and comprehensible manner, contributing to the adoption of i* framework in the industry. The proposed approach is designed by considering the semantics of i* constructs and layout conventions of i* models in order to produce meaningful layouts and can appropriately handle both the SD (Strategic Dependency) view and the SR (Strategic Rationale) view of i* models. We have implemented our approach in an open-access prototype tool, which is able to be integrated with existing iStarML-compatible modeling tools. We have conducted a controlled experiment, a case study, and performance testing to empirically and comprehensively evaluate the utility of our approach, the results of which show that our proposal can efficiently produce meaningful layouts that are as comprehensible as manually laid out models in most cases.
... The reasons were twofold. First, communities may overlap, which could have introduced visual ambiguities in community distinguishing [79]. Second, the community-related objective questions mainly inquired about the exact or approximate number of communities in a graph, and the corresponding background stories implied the answers. ...
A graph is an abstract model that represents relations among entities, for example, the interactions of characters in a novel. Background story endows entities and relations with real-world meanings and describes semantics and contexts of the abstract model, for example, the actual story that the novel presents. Considering practical experience and relevant research, human viewers who know the background story of a graph and those not knowing the story may perform differently when perceiving the same graph. However, there are currently no previous studies to adequately address this problem. This paper presents an evaluation study that investigates the effects of background stories on graph perception. We formulate three hypotheses on different aspects including visual focus areas, graph structure identification, and mental model formation, and design three controlled experiments to test our hypotheses using real-world graphs with background stories. We analyze our experimental data to compare the performance of participants who have read and not read the background stories, and obtain a set of instructive findings. First, our results show that knowing the background stories affects participants focus areas in interactive graph exploration to a certain extent. Second, it significantly affects the performance of identifying community structures but not high degree and bridge structures. Third, it has a significant impact on graph recognition under blurred visual conditions. These findings can bring new considerations to the design of storytelling visualizations and interactive graph explorations.
... The reasons were twofold. First, communities may overlap, which could have introduced visual ambiguities in community distinguishing [79]. Second, the community-related objective questions mainly inquired about the exact or approximate number of communities in a graph, and the corresponding background stories implied the answers. ...
A graph is an abstract model that represents relations among entities, for example, the interactions between characters in a novel. A background story endows entities and relations with real-world meanings and describes the semantics and context of the abstract model, for example, the actual story that the novel presents. Considering practical experience and prior research, human viewers who are familiar with the background story of a graph and those who do not know the background story may perceive the same graph differently. However, no previous research has adequately addressed this problem. This research paper thus presents an evaluation that investigated the effects of background stories on graph perception. Three hypotheses that focused on the role of visual focus areas, graph structure identification, and mental model formation on graph perception were formulated and guided three controlled experiments that evaluated the hypotheses using real-world graphs with background stories. An analysis of the resulting experimental data, which compared the performance of participants who read and did not read the background stories, obtained a set of instructive findings. First, having knowledge about a graph's background story influences participants' focus areas during interactive graph explorations. Second, such knowledge significantly affects one's ability to identify community structures but not high degree and bridge structures. Third, this knowledge influences graph recognition under blurred visual conditions. These findings can bring new considerations to the design of storytelling visualizations and interactive graph explorations.
... Many of these techniques are unsupervised or semi-supervised, so cluster boundaries and node memberships can be uncertain. Some interactive visualization tools communicate this uncertainty by showing crisp and overlapping community structures [86,87] to address fuzzy overlapping communities with different granularity. ...
Probabilistic graphs are challenging to visualize using the traditional node-link diagram. Encoding edge probability using visual variables like width or fuzziness makes it difficult for users of static network visualizations to estimate network statistics like densities, isolates, path lengths, or clustering under uncertainty. We introduce Network Hypothetical Outcome Plots (NetHOPs), a visualization technique that animates a sequence of network realizations sampled from a network distribution defined by probabilistic edges. NetHOPs employ an aggregation and anchoring algorithm used in dynamic and longitudinal graph drawing to parameterize layout stability for uncertainty estimation. We present a community matching algorithm to enable visualizing the uncertainty of cluster membership and community occurrence. We describe the results of a study in which 51 network experts used NetHOPs to complete a set of common visual analysis tasks and reported how they perceived network structures and properties subject to uncertainty. Participants' estimates fell, on average, within 11% of the ground truth statistics, suggesting NetHOPs can be a reasonable approach for enabling network analysts to reason about multiple properties under uncertainty. Participants appeared to articulate the distribution of network statistics slightly more accurately when they could manipulate the layout anchoring and the animation speed. Based on these findings, we synthesize design recommendations for developing and using animated visualizations for probabilistic networks.
