IVAPP 2023 Abstracts

Area 1 - General Visualization Techniques
Full Papers Short Papers
Area 2 - Information Visualization
Full Papers Short Papers
Area 3 - Scientific Visualization
Short Papers

Area 1 - General Visualization Techniques

Full Papers
Paper Nr: 11
Title:

Model Order in Sugiyama Layouts
Authors:

Sören Domrös, Max Riepe and Reinhard von Hanxleden
Abstract: Graph drawing algorithms traditionally consider a graph to consist of unordered sets of nodes and edges, which may disregard information already provided by the developer. In practice, as recently argued by (Domrös and von Hanxleden, 2022), a graph often consists of ordered sets, which have an intended model order of nodes and edges. We present how this model order can be enforced or used as a tie-breaker, while optimizing common aesthetic criteria. This allows the developer to control the layout of layered graphs via the model order. On the example of SCCharts, we show that the order of nodes and edges does indeed correlate with the way people think about a model, and how that order can be used to emphasize the semantics of a sensibly designed model. Moreover, we suggest model order strategies to be used for control-flow and data-flow diagrams based on expert developer feedback on SCCharts and Lingua Franca.
Download
Paper Nr: 36
Title:

A Survey of Geospatial-Temporal Visualizations for Military Operations
Authors:

G. Walsh, N. S. Andersen, N. Stoianov and S. Jänicke
Abstract: For the first time European defence funding has surpassed 200 billion euros per year, with a renewed strategic interest in creating technological innovations which aid military co-operation such as comprehensive command and control information systems. Overcoming the many challenges associated with the research and development of such military technologies presents an excellent opportunity for the visualization community’s contributions in the domain as there is ample scope for applied research. This survey is interested in further developing the functionality of military decision-support systems by assessing the integration of cutting-edge geospatial-temporal visualizations into such systems. With this objective in mind, this survey systematically identifies, investigates, and discusses suitable visualization solutions and the benefit they may offer to military command and control systems through the lens of the Military Operations Process. The survey identifies gaps and opportunities for improvement of existing military products where identified geospatial-temporal visualizations can enhance military commanders decision-making capabilities and their ability to act. No other recent surveys examine such information visualizations and visual analytics tools used in the military domain. The survey results in the formulation of a design space and guidelines to be used in the design process of visualization and visual analytics tools supporting military operations, based on the assessment of a visualizations relevance and characteristics according to each phase of the Military Operations Process.
Download
Short Papers
Paper Nr: 12
Title:

Evaluating Differences in Insights from Interactive Dimensionality Reduction Visualizations Through Complexity and Vocabulary
Authors:

Mia Taylor, Lata Kodali, Leanna House and Chris North
Abstract: The software, Andromeda, enables users to explore high-dimensional data using the dimensionality reduction algorithm Weighted Multidimensional Scaling (WMDS). How data are projected in WMDS is determined by weights assigned to variables, and with Andromeda, the weights are set in response to user interactions. This work evaluates the impact of such interactions on student insight generation via a large-scale study implemented in a university introductory statistics course. Insights are analyzed using complexity metrics. This analysis is extended to compare insight vocabulary to gain an understanding of differences in terminology. Both analyses are conducted using the same semi-automated method that applies basic natural language processing techniques and logistic regression modeling. Results show that specific user interactions correlate to differences in the dimensionality and cardinality of insights. Overall, these results suggest that the interactions available to users impact their insight generation and therefore impact their learning and analysis process.
Download
Paper Nr: 19
Title:

XAIVIER the Savior: A Web Application for Interactive Explainable AI in Time Series Data
Authors:

Ilija Šimić, Christian Partl and Vedran Sabol
Abstract: The rising popularity of black-box deep learning models directly lead to an increased interest in eXplainable AI - a field concerned with methods that explain the behavior of machine learning models. However, different types of stakeholders interact with XAI, all of which have different requirements and expectations of XAI systems. Moreover, XAI methods and tools are mostly developed for image, text, and tabular data, while explainability methods and tools for time series data - which is abundant in high-stakes domains - are in comparison fairly neglected. In this paper, we first contribute with a set of XAI user requirements for the most prominent XAI stakeholders, the machine learning experts. We also contribute with a set of functional requirements, which should be fulfilled by an XAI tool to address the derived user requirements. Based on the functional requirements, we have designed and developed XAIVIER, the eXplainable AI VIsual Explorer and Recommender, a web application for interactive XAI in time series data. XAIVIER stands out with its explainer recommender that advises users which explanation method they should use for their dataset and model, and which ones to avoid. We have evaluated XAIVIER and its explainer recommender in a usability study, and demonstrate its usage and benefits in a detailed user scenario.
Download
Paper Nr: 30
Title:

On Metavisualization and Properties of Visualization
Authors:

Jaya Sreevalsan-Nair
Abstract: Metavisualization is the “visualization of visualizations” which is the commonly used definition. However, there is a gap in the theoretical foundations of metavisualization. This gap has led to the under-utilization of metavisualization, which is much needed today, given the proliferation of the use of visualizations in data science. Two observations have inspired this work to build the theory of metavisualization: (i) the interdisciplinary differences in the understanding of metavisualization, and (ii) the inter-relationships between metavisualization, analysis of visualizations, and visual analytics. Hence, we conduct a systematic literature review on metavisualization, identify visualization properties that can be used for generating a metavisualization, and propose a design space for these properties. This work is a theoretical discourse on metavisualization of visualization and its properties, based on the visualization-based understanding and practice of metavisualization.
Download
Paper Nr: 31
Title:

An Interactive Graph Layout Constraint Framework
Authors:

Jette Petzold, Sören Domrös, Connor Schönberner and Reinhard von Hanxleden
Abstract: Several solutions exist to constrain nodes or edges for creating desired graph layouts or arrangements via automatic layout. However, these constraints, which are often handled in separate views, tend to produce conflicts if not handled carefully. We present an interactive layout framework that visualizes existing constraints and available new constraints interactively in the diagram. Adding constraints via diagram interaction allows reevaluation of existing constraints based on the intended movement of the constrained node and prevents conflicts between constraints. The framework can easily be utilized by new layout algorithms and is independent of the actual layout implementation.
Download

Area 2 - Information Visualization

Full Papers
Paper Nr: 2
Title:

Visualizing Grassmannians via Poincare Embeddings
Authors:

Huanran Li and Daniel Pimentel-Alarcón
Abstract: This paper introduces an embedding to visualize high-dimensional Grassmannians on the Poincaré disk, obtained by minimizing the KL-divergence of the geodesics on each manifold. Our main theoretical result bounds the loss of our embedding by a log-factor of the number of subspaces, and a term that depends on the distribution of the subspaces in the Grassmannian. This term will be smaller if the subspaces form well-defined clusters, and larger if the subspaces have no structure whatsoever. We complement our theory with synthetic and real data experiments showing that our embedding can provide a more accurate visualization of Grassmannians than existing representations.
Download
Paper Nr: 3
Title:

Damast: A Visual Analysis Approach for Religious History Research
Authors:

Max Franke and Steffen Koch
Abstract: Digital humanities (DH) combines research objectives in the humanities with digital data acquisition, processing, and presentation methods. This work describes the development of a visualization approach in the field of DH to analyze the coexistence of institutionalized religious communities in Middle Eastern cities during the medieval period. Our approach aims to support the entire process of data acquisition, storage, analysis, and publication with interactive visualization. The support of the whole process enables a consistent concept for the representation of confidence, the collection of provenance information, and the implicit storage of gained knowledge. Our concept empowers scholars to trace obtained results up to the verifiability of details in the corresponding sources, facilitates collaborative analyses, and allows for the serialization of results and use in corresponding publications. We also reflect on the benefits, limitations, and lessons learned when applying interactive visualization to the concrete tasks and with respect to data collection and publication of findings.
Download
Paper Nr: 5
Title:

Contrast Driven Color-Group Assignment in Categorical Data Visualization
Authors:

Éric Languenou
Abstract: Ubiquitous digital technology has facilitated the collect of multi-dimensional numerical data that are analyzed by specialists. Their need to explore and to explain this data to non-specialists is important. With categorical data, we construct various diagrams on a color-coded paradigm associating colors with data classes. Depending on the number of classes or the geometry of diagrams, the class-color assignment choice can become a complicated task, with the number of permutations growing in a factorial way with the number of categories. The goal of this research is to develop an algorithm aiming at assigning the best color, among a user given color palette, for each class of objects of a categorical data visualization. We optimize the ability, for a viewer, to distinguish classes’ geometrical objects one from another using a concept of contrast importance factors expressing the need to get for a pair of objects classes a high color contrast. The method relies on a fitness function separation between palette color distances and geometrical contrast need. We indicate applications of the concept to two kinds of categorical visualizations: streamgraphs and chord diagrams for which optimized color assignment has never been published so far.
Download
Paper Nr: 10
Title:

Viewpoint-Based Quality for Analyzing and Exploring 3D Multidimensional Projections
Authors:

Wouter Castelein, Zonglin Tian, Tamara Mchedlidze and Alexandru Telea
Abstract: While 2D projections are established tools for exploring high-dimensional data, the effectiveness of their 3D counterparts is still a matter of debate. In this work, we address this from a multifaceted quality perspective. We first propose a viewpoint-dependent definition of 3D projection quality and show how this captures the visual variability in 3D projections much better than aggregated, single-value, quality metrics. Next, we propose an interactive exploration tool for finding high-quality viewpoints for 3D projections. We use our tool in an user evaluation to gauge how our quality metric correlates with user-perceived quality for a cluster identification task. Our results show that our metric can predict well viewpoints deemed good by users and that our tool increases the users’ preference for 3D projections as compared to classical 2D projections.
Download
Paper Nr: 13
Title:

Using Well-Known Techniques to Visualize Characteristics of Data Quality
Authors:

Roy A. Ruddle
Abstract: Previous work has identified more than 100 distinct characteristics of data quality, most of which are aspects of completeness, accuracy and consistency. Other work has developed new techniques for visualizing data quality, but there is a lack of research into how users visualize data quality issues with existing, well-known techniques. We investigated how 166 participants identified and illustrated data quality issues that occurred in a 54-file, longitudinal collection of open data. The issues that participants identified spanned 27 different characteristics, nine of which do not appear in existing data quality taxonomies. Participants adopted nine visualization and tabular methods to illustrate the issues, using the methods in five ways (quantify; alert; examples; serendipitous discovery; explain). The variety of serendipitous discoveries was noteworthy, as was how rarely participants used visualization to illustrate completeness and consistency, compared with accuracy. We conclude by presenting a 106-item data quality taxonomy that combines seven previous works with our findings.
Download
Paper Nr: 14
Title:

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity
Authors:

Fairouz Grioui and Tanja Blascheck
Abstract: We investigate three visualizations showing heart rate (HR) and HR zones (HRZ) data collected over time and displayed on a virtual smartwatch, to monitor physical activity intensity. To understand exercise behavior, we first conducted a survey with 57 participants and found that most of them track their activities (66%) using wrist wearable devices (i. e., smartwatches or fitness bands) and that during the course of the exercise data is primarily represented as text or a combination of text and icon. To support reaching a specific fitness goal, we designed a bar chart visualization combining both current and historical HR and HRZ data. Among the three visualizations, two present an additional chart (i. e., a horizontal and radial bar chart summary), showing the amount of time spent per HRZ (i. e., low, moderate, and high intensity). In a controlled study performed in virtual reality, we compared participants’ performance with each visualization asking participants to make a quick and accurate decision while exercising (i. e., playing a tennis-like game). Results from the study show evidence of a difference in task performance between visualizations with and without a summary chart—visualizations showing a summary chart performed better than the version without. Finally, based on our study results we present lessons learned.
Download
Short Papers
Paper Nr: 6
Title:

Visual Document Exploration with Adaptive Level of Detail: Design, Implementation and Evaluation in the Health Information Domain
Authors:

L. Shao, S. Lengauer, H. Miri, M. A. Bedek, B. Kubicek, C. Kupfer, M. Zangl, B. C. Dienstbier, K. Jeitler, C. Krenn, T. Semlitsch, C. Zipp, D. Albert, A. Siebenhofer and T. Schreck
Abstract: Documents typically show a linear structure in which the content can be accessed. However, linear reading is not always desired by users, nor is it the best presentation way, as information needs may be developing or changing over time, and users would thus want to extract the relevant information by navigation and search. Therefore, reading with adaptive focus and level of detail is needed. This is of utmost importance in the health information domain where patient conditions and resulting information needs may evolve in different directions over time. We report on the development of a visual document exploration system which supports navigating a document at different levels of aggregation, from topic overview (high-level) to keyword occurrences (mid-level) to full text (low-level). Our design smoothly integrates the different levels of detail from which the users can choose. The system is designed to track explored topics and use this information to suggest additional content. We evaluated the design and its corresponding web-based implementation through a formative user-study in the domain of diabetes health information. The evaluation confirmed that our design and implementation can raise interest and curiosity, and also allow users to efficiently navigate content of interest.
Download
Paper Nr: 8
Title:

The Compilation of 2D and 3D Dynamic Visualizations
Authors:

Brian Farrimond and Ella Pereira
Abstract: 3D modelling and visualization are rapidly developing in power and application. Unfortunately they are also developing in complexity of use. They require considerable practice and skill in order to model and visualize successfully. This paper presents modelling and visualization strategies and tools based on textual descriptions of models and visualizations. The principles of compilation used in coding for many decades are applied to modelling and visualization. This results in tools able to model and visualize many types of dynamic object such as ships and locomotives that can be used successfully by non-expert users who have knowledge of the objects being modelled. The tools have been used in local primary schools since 2007.
Download
Paper Nr: 20
Title:

Towards a Visual Analytics Workflow for Cybersecurity Simulations
Authors:

Vit Rusnak and Martin Drasar
Abstract: One of the contemporary grand challenges in cybersecurity research is designing and evaluating effective attack strategies on network infrastructures performed by autonomous agents. These attackers are developed and trained in simulated environments. While the simulation environments are maturing, their support for analyzing the simulation data remains limited, mainly to inspect individual simulation runs. Extending the analytical workflow to compare multiple runs and integrating visualizations could improve the design of both attack and defense strategies. Through our work, we want to spark interest in the largely overlooked domain of visual analytics for cybersecurity simulation workflows. In this paper, we a) analyze the current state of the art of using visualizations in cybersecurity simulations; b) conceptualize the three-tier analytical workflow and identify user tasks with suggested visualizations for each tier; c) demonstrate the use of visualizations that augment existing CYST simulator on several real-world tasks and discuss the limitations and lessons learned.
Download
Paper Nr: 22
Title:

Evaluating Architectures and Hyperparameters of Self-supervised Network Projections
Authors:

Tim Cech, Daniel Atzberger, Willy Scheibel, Rico Richter and Jürgen Döllner
Abstract: Self-Supervised Network Projections (SSNP) are dimensionality reduction algorithms that produce low-dimensional layouts from high-dimensional data. By combining an autoencoder architecture with neighborhood information from a clustering algorithm, SSNP intend to learn an embedding that generates visually separated clusters. In this work, we extend an approach that uses cluster information as pseudo-labels for SSNP by taking outlier information into account. Furthermore, we investigate the influence of different autoencoders on the quality of the generated two-dimensional layouts. We report on two experiments on the autoencoder’s architecture and hyperparameters, respectively, measuring nine metrics on eight labeled datasets from different domains, e.g., Natural Language Processing. The results indicate that the model’s architecture and the choice of hyperparameter values can influence the layout with statistical significance, but none achieves the best result over all metrics. In addition, we found out that using outlier information for the pseudo-labeling approach can maintain global properties of the two-dimensional layout while trading-off local properties.
Download
Paper Nr: 26
Title:

BigGraphVis: Visualizing Communities in Big Graphs Leveraging GPU-Accelerated Streaming Algorithms
Authors:

Ehsan Moradi and Debajyoti Mondal
Abstract: Graph layouts are key to exploring massive graphs. Motivated by the advances in streaming community detection methods that process the edge list in one pass with only a few operations per edge, we examine whether they can be leveraged to rapidly create a coarse visualization of the graph communities, and if so, then how the quality would compare with the layout of the whole graph. We introduce BigGraphVis which combines a parallelized streaming community detection algorithm and probabilistic data structure to leverage the parallel processing power of GPUs to visualize graph communities. To the best of our knowledge, this is the first attempt to combine the potential of streaming algorithms coupled with GPU computing to tackle community visualization challenges in big graphs. Our method extracts community information in a few passes on the edge list, and renders the community structures using a widely used ForceAtlas2 algorithm. The coarse layout generation process of BigGraphVis is 70 to 95 percent faster than computing a GPU-accelerated ForceAtlas2 layout of the whole graph. Our experimental results show that BigGraphVis can produce meaningful layouts, and thus opens up future opportunities to design streaming algorithms that achieve a significant computational speed up for massive networks by carefully trading off the layout quality.
Download
Paper Nr: 27
Title:

The HORM Diagramming Tool: A Domain-Specific Modelling Tool for SME Cybersecurity Awareness
Authors:

Costas Boletsis, Sefat N. Orni and Ragnhild Halvorsrud
Abstract: Improving security posture while addressing human errors made by employees are among the most challenging tasks for SMEs concerning cybersecurity risk management. To facilitate these measures, a domain-specific modelling tool for visualising cybersecurity-related user journeys, called the HORM Diagramming Tool (HORM-DT), is introduced. By visualising SMEs’ cybersecurity practices, HORM-DT aims to raise their cybersecurity awareness by highlighting the related gaps, thereby ultimately informing new or updated cyber-risk strategies. HORM-DT’s target group consists of SMEs’ employees with various areas of technical expertise and different backgrounds. The tool was developed as part of the Human and Organisational Risk Modelling (HORM) framework, and the underlying formalism is based on the Customer Journey Modelling Language (CJML) as extended by elements of the CORAS language to cover cybersecurity-related user journeys. HORM-DT is a fork of the open-source Diagrams.net software, which was modified to facilitate the creation of cybersecurity-related diagrams. To evaluate the tool, a usability study following a within-subject design was conducted with 29 participants. HORM-DT achieved a satisfactory system usability scale score of 80.69, and no statistically significant differences were found between participants with diverse diagramming tool experience. The tool’s usability was also praised by participants, although there were negative comments regarding its functionality of connecting elements with lines.
Download
Paper Nr: 28
Title:

Interactive Exploration of Complex Heterogeneous Data: A Use Case on Understanding City Economics
Authors:

Rainer Splechtna, Thomas Hulka, Disha Sardana, Nikitha D. Chandrashekar, Denis Gračanin and Krešimir Matković
Abstract: The analysis of complex, heterogeneous data containing spatial and temporal components is a non-trivial task. Besides data heterogeneity and data quantity, the exploratory nature of data analysis tasks, which is only roughly specified at the beginning and refined during the process, poses main challenges. In this paper, we describe a holistic approach to interactive visual analysis of such data. We use the IEEE VAST Challenge 2022 data set for this purpose. To support the exploratory tasks dealing with the economic health of a city, we apply different data processing, introduce new views, and employ complex interactions. All these steps are necessary for an efficient workflow. We rely on the well-known paradigm of coordinated multiple views. In addition to the standard views, we introduce the interactive map view, which supports the visualization of different statistical values on the map itself. All views are interactive and support multiple composite brushing. Our results, which are also described, illustrate the effectiveness of our approach and show its applicability to similar data and tasks.
Download
Paper Nr: 32
Title:

Supporting University Research and Administration via Interactive Visual Exploration of Bibliographic Data
Authors:

Kostiantyn Kucher and Andreas Kerren
Abstract: Bibliographic data and bibliometric analyses play an important role in the professional life of academic researchers, and the quality of the respective publication records is essential for establishing the big picture of the relationships between particular publications, their authors and affiliations, or further data facets associated with publications. In this paper, we report on the design and outcomes of an interactive visual data exploration project conducted within the scope of a university with the goal of gaining overview of the university publication data. The project has been carried out by information visualization researchers in collaboration with several groups of stakeholders, including the university library and administration staff. We describe the design considerations, the resulting interactive visual interface, and the feedback received from the stakeholders with respect to the tool functionality and the insights discovered in the bibliographic data.
Download
Paper Nr: 34
Title:

Visual Analysis of Multi-Labelled Temporal Noise Data from Multiple Sensors
Authors:

Juan José Franco and Pere-Pau Vázquez
Abstract: Environmental noise pollution is a problem for cities’ inhabitants, that can be especially severe in large cities. To implement measures that can alleviate this problem, it is necessary to understand the extent and impact of different noise sources. Although gathering data is relatively cheap, processing and analyzing the data is still complex. Besides the lack of an automatic method for labelling city sounds, maybe more important is the fact that there is not a tool that allows domain experts to analytically explore data that has been manually labelled. To solve this problem, we have created a visual analytics application that facilitates the exploration of multiple-labelled temporal data captured at four different corners of a crossing in a populated area of Barcelona, the Eixample neighborhood. Our tool consists of a series of linked interactive views that facilitate top-down (from noise events to labels) and bottom-up (from labels to time slots) exploration of the captured data.
Download
Paper Nr: 7
Title:

Trajectory-Based Dynamic Boundary Map Labeling
Authors:

Ming-Hsien Wu and Hsu-Chun Yen
Abstract: Traditional map labeling focuses on placing labels on a static map to help the reader gain a better understanding of the content of the map. As the content of a dynamic map changes as time progresses, traditional static map labeling algorithms usually cannot be applied to dynamic maps directly. In this paper, we consider the design of algorithms for trajectory-based dynamic boundary labeling, in which non-overlapping labels, connecting to points on the map through straight-line leaders, are placed on one side of a viewing window which moves or rotates along a trajectory. The goal is to maximize the total visible time of all labels during the course of the navigation. To avoid visual disruptions, the effect of flickering is also taken into account in our design. Even though the problem can be formulated using mathematical optimization, heuristic strategies are also incorporated in the design to reduce the running time to make the solutions more practical in real-world applications. Finally, experimental results are used to illustrate the effectiveness of our design.
Download
Paper Nr: 29
Title:

A Comparative Study on Vision Transformers in Remote Sensing Building Extraction
Authors:

Georgios-Fotios Angelis, Armando Domi, Alexandros Zamichos, Maria Tsourma, Ioannis Manakos, Anastasios Drosou and Dimitrios Tzovaras
Abstract: Data visualization has received great attention in the last few years and gives valuable assets for better understanding and extracting information from data. More specifically, in Geospatial data, visualization includes information about the location, the geometric shape of elements, and the exact position of elements that can lead in enhances downstream applications such as damage detection, building energy consumption estimation, urban planning and change detection. Extracting building footprints from remote sensing (RS) imagery can help in visualizing damaged buildings and separate them form terrestrial objects. Considering this, the current manuscript provides a detailed comparison and a new benchmark for remote sensing building extraction. Experiments are conducted in three publicly available datasets aiming to evaluate accuracy and performance of the compared Transformer-based architectures. MiTNet and other five transformers architectures are introduced, namely DeepViTUNet, DeepViTUNet++, Coordformer, PoolFormer, EfficientFormer. In these choices we study design adjustments in order to obtain the best trade off between computational cost and performance. Experimental findings demonstrate that MitNet, which learns features in a hierarchical manner can be established as a new benchmark.
Download

Area 3 - Scientific Visualization

Short Papers
Paper Nr: 35
Title:

MR to CT Synthesis Using GANs: A Practical Guide Applied to Thoracic Imaging
Authors:

Arthur Longuefosse, Baudouin Denis De Senneville, Gaël Dournes, Ilyes Benlala, François Laurent, Pascal Desbarats and Fabien Baldacci
Abstract: In medical imaging, MR-to-CT synthesis has been extensively studied. The primary motivation is to benefit from the quality of the CT signal, i.e. excellent spatial resolution, high contrast, and sharpness, while avoiding patient exposure to CT ionizing radiation, by relying on the safe and non-invasive nature of MRI. Recent studies have successfully used deep learning methods for cross-modality synthesis, notably with the use of conditional Generative Adversarial Networks (cGAN), due to their ability to create realistic images in a target domain from an input in a source domain. In this study, we examine in detail the different steps required for cross-modality translation using GANs applied to MR-to-CT lung synthesis, from data representation and pre-processing to the type of method and loss function selection. The different alternatives for each step were evaluated using a quantitative comparison of intensities inside the lungs, as well as bronchial segmentations between synthetic and ground truth CTs. Finally, a general guideline for cross-modality medical synthesis is proposed, bringing together best practices from generation to evaluation.
Download