IVAPP 2022 Abstracts

Area 1 - Abstract Data Visualization
Full Papers Short Papers
Area 2 - General Data Visualization
Full Papers Short Papers
Area 3 - Spatial Data Visualization
Full Papers Short Papers

Area 1 - Abstract Data Visualization

Full Papers
Paper Nr: 2
Title:

Interactive Input and Visualization for Planning with Temporal Uncertainty
Authors:

M. Höhn, M. Wunderlich, K. Ballweg, J. Kohlhammer and T. von Landesberger
Abstract: Data with temporal uncertainty is ubiquitous in everyone’s life. Popular examples are holiday planning or train trips. There are several approaches to visualize temporal uncertainty, but common research usually does not take uncertainty into account, neither as input nor output. We propose a new approach that provides both an interactive drawing for data with temporal uncertainty and their respective visualizations. The user can draw both variable and fixed activities and also has the possibility to set probability distributions and enter indefinite activities. A quantitative user study shows the need and suitability of our new approach.
Download
Paper Nr: 4
Title:

Probabilistic Envelope based Visualization for Monitoring Drilling Well Data Logging
Authors:

Kishansingh Rajput and Guoning Chen
Abstract: In oil and gas industries, to monitor the drilling well status and take actions when needed to prevent damage, different logs are recorded and compared with the reference logs of the nearby existing wells. The deviation of the log of the current well from the majority of the reference logs may indicate potential issues of drilling. Currently, the standard methods used in the industry are line/scatter plots. Due to limitations such as clutter and lack of quantitative information, these plots are not effective. In this paper, a probabilistic envelope based technique is proposed for the visualization and anomaly detection of drilling data. This technique provides quantitative information, is able to avoid the outliers in the reference data and works well even with a large number of reference sequences. This technique is applied to the detection of anomalies from drilling well logs to demonstrate its effectiveness. It is also adapted to the detection of over/under gauge during drilling.
Download
Paper Nr: 14
Title:

SDR-NNP: Sharpened Dimensionality Reduction with Neural Networks
Authors:

Youngjoo Kim, Mateus Espadoto, Scott C. Trager, Jos M. Roerdink and Alexandru C. Telea
Abstract: Dimensionality reduction (DR) methods aim to map high-dimensional datasets to 2D scatterplots for visual exploration. Such scatterplots are used to reason about the cluster structure of the data, so creating well-separated visual clusters from existing data clusters is an important requirement of DR methods. Many DR methods excel in speed, implementation simplicity, ease of use, stability, and out-of-sample capabilities, but produce suboptimal cluster separation. Recently, Sharpened DR (SDR) was proposed to generically help such methods by sharpening the data-distribution prior to the DR step. However, SDR has prohibitive computational costs for large datasets. We present SDR-NNP, a method that uses deep learning to keep the attractive sharpening property of SDR while making it scalable, easy to use, and having the out-of-sample ability. We demonstrate SDR-NNP on seven datasets, applied on three DR methods, using an extensive exploration of its parameter space. Our results show that SDR-NNP consistently produces projections with clear cluster separation, assessed both visually and by four quality metrics, at a fraction of the computational cost of SDR. We show the added value of SDR-NNP in a concrete use-case involving the labeling of astronomical data.
Download
Paper Nr: 23
Title:

SDBM: Supervised Decision Boundary Maps for Machine Learning Classifiers
Authors:

Artur M. Oliveira, Mateus Espadoto, Roberto Hirata Jr. and Alexandru C. Telea
Abstract: Understanding the decision boundaries of a machine learning classifier is key to gain insight on how classifiers work. Recently, a technique called Decision Boundary Map (DBM) was developed to enable the visualization of such boundaries by leveraging direct and inverse projections. However, DBM have scalability issues for creating fine-grained maps, and can generate results that are hard to interpret when the classification problem has many classes. In this paper we propose a new technique called Supervised Decision Boundary Maps (SDBM), which uses a supervised, GPU-accelerated projection technique that solves the original DBM shortcomings. We show through several experiments that SDBM generates results that are much easier to interpret when compared to DBM, is faster and easier to use, while still being generic enough to be used with any type of single-output classifier.
Download
Short Papers
Paper Nr: 7
Title:

NetVisGame: Mobile Gamified Information Visualization of Home Network Traffic Data
Authors:

Marija Schufrin, Katharina Kuban, Arjan Kuijper and Jörn Kohlhammer
Abstract: The awareness of everyday internet users for cyber security becomes ever more important considering the ubiquity of the Internet in everyday life. However, people usually lack the necessary understanding of this topic or the motivation to pay attention to the problem and its possible consequences. In this paper, we present the novel idea of combining visualization of one’s own personal data related to cyber-security literacy with a casual gaming approach. We therefore introduce our prototype, NetVisGame, in which we have implemented the idea for personal network traffic data based on a preliminary user study. The evaluation results of the first iteration of the user-centered design process supports the assumption that this approach is feasible to raise interest for and foster understanding of network traffic data and therefore could be a promising approach for data and technologies related to cyber-security literacy.
Download
Paper Nr: 17
Title:

Preserving Order during Crossing Minimization in Sugiyama Layouts
Authors:

Sören Domrös and Reinhard Von Hanxleden
Abstract: The Sugiyama algorithm, also known as the layered algorithm or hierarchical algorithm, is an established algorithm to produce crossing-minimal drawings of graphs. It does not, however, consider an initial order of the vertices and edges. We show how ordering real vertices, dummy vertices, and edge ports before crossing minimization may preserve the initial order given by the graph without compromising, on average, the quality of the drawing regarding edge crossings. Even for solutions in which the initial graph order produces more crossings than necessary or the vertex and edge order is conflicting, the proposed approach can produce better crossing-minimal drawings than the traditional approach.
Download
Paper Nr: 33
Title:

Visualization of Knowledge Distribution across Development Teams using 2.5D Semantic Software Maps
Authors:

Daniel Atzberger, Tim Cech, Adrian Jobst, Willy Scheibel, Daniel Limberger, Matthias Trapp and Jürgen Döllner
Abstract: In order to detect software risks at an early stage, various software visualization techniques have been developed for monitoring the structure, behaviour, or the underlying development process of software. One of greatest risks for any IT organization consists in an inappropriate distribution of knowledge among its developers, as a projects’ success mainly depends on assigning tasks to developers with the required skills and expertise. In this work, we address this problem by proposing a novel Visual Analytics framework for mining and visualizing the expertise of developers based on their source code activities. Under the assumption that a developer’s knowledge about code is represented directly through comments and the choice of identifier names, we generate a 2D layout using Latent Dirichlet Allocation together with Multidimensional Scaling on the commit history, thus displaying the semantic relatedness between developers. In order to capture a developer’s expertise in a concept, we utilize Labeled LDA trained on a corpus of Open Source projects. By mapping aspects related to skills onto the visual variables of 3D glyphs, we generate a 2.5D Visualization, we call KnowhowMap. We exemplify this approach with an interactive prototype that enables users to analyze the distribution of skills and expertise in an explorative way.
Download
Paper Nr: 24
Title:

Software Architecture Mining from Source Code with Dependency Graph Clustering and Visualization
Authors:

Anthony Savidis and Crystallia Savaki
Abstract: The software architecture represents an important asset, constituting a shared vision amongst the software engineers of the various system components. Good architectures link to modular design, with loose coupling and cohesion defining which operations are grouped together to form a modular architectural entity. Modularity is achieved by practice otherwise we may observe a mismatch where the source code diverges from the primary architectural vision. In fact, class groups with dense interdependencies denote the real architectural entities as derived and implied directly from source code. In this work, we created a tool to assist in mining the actual system architecture. We extract all sorts of dependencies by processing all source files, and then using graph clustering, we capture and interactively visualize strongly coupled class groups with configurable weights. We also support forced clustering on namespaces, packages and folders.
Download
Paper Nr: 28
Title:

Task-based Evaluation of Sentiment Visualization Techniques
Authors:

Kostiantyn Kucher, Samir Bouchama, Achim Ebert and Andreas Kerren
Abstract: Sentiment visualization is concerned with visual representation of sentiments, emotions, opinions, and stances typically detected in textual data, for example, charts or diagrams representing negative and positive opinions in online customer reviews or Twitter discussions. Such approaches have been applied for the purposes of academic research and practical applications in the past years. But the question of usability of these various techniques still remains generally unsolved, as the existing research typically addresses individual design alternatives for a particular technique implementation only. This work focuses on evaluation of the effectiveness and efficiency of common visual representations for low-level visualization tasks in the context of sentiment visualization. More specifically, we describe a task-based within-subject user study for various tasks, carried out as an online survey and taking the task completion time and error rate into account for most questions. The study involved 50 participants, and we present and discuss their responses and free-form comments. The results provide evidence of strengths and weaknesses of particular representations and visual variables with respect to different tasks, as well as specific user preferences, in the context of sentiment visualization.
Download
Paper Nr: 32
Title:

Semi-automatic CNN Architectural Pruning using the Bayesian Case Model and Dimensionality Reduction Visualization
Authors:

Wilson E. Marcílio-Jr., Danilo M. Eler and Ivan R. Guilherme
Abstract: Visualization techniques have been applied to reasoning about complex machine learning models. These visual approaches aim to enhance the understanding of black-box models’ decisions or guide in hyperparameters configuration, such as the number of layers and neurons/filters in deep neural networks. While several works address the architectural tuning of convolutional neural networks (CNNs), only a few works face the problem from a semi-automatic perspective. This work presents a novel application of the Bayesian Case Model that uses visualization strategies to convey the most important filters of convolutional layers for image classification. A heatmap coordinated with a scatterplot visualization emphasizes the filters with the most contribution to the CNN prediction. Our methodology is evaluated on a case study using the MNIST dataset.
Download

Area 2 - General Data Visualization

Full Papers
Paper Nr: 3
Title:

ConText: Supporting the Pursuit and Management of Evidence in Text-based Reporting Systems
Authors:

Tabassum Kakar, Xiao Qin, Elke Rundensteiner, Lane Harrison, Sanjay Sahoo, Suranjan De and Thang La
Abstract: Instance-based Incident Analysis (IIA) – a labor intensive and error-prone task – requires analysts to review text-based reports of incidents, where each may be evidence of a larger problem that requires regulatory action. Given the rise of reporting systems in many organizations, there is a need to explore tools that may aid IIA analysts in exploring, evaluating, and generalizing findings across a large set of independently produced reports in a unified workflow – currently not supported by existing tools. In this work, we contribute a design study conducted in collaboration with Pharmacovigilance experts at the US Food and Drug Administration. Following a series of interviews and discussions focused on workflows and toolsets, we develop a prototype, ConText, which combines domain-informed computational methods with interactive visual displays to support evidence identification, collection, and management for IIA. We evaluate ConText via case studies and follow-up semi-structured interviews, depicting its effectiveness in performing IIA tasks of evidence collection and monitoring. We discuss insights derived from the design and evaluation of ConText that may be valuable for designing future interactive analytic systems for life-critical IIA workflows.
Download
Short Papers
Paper Nr: 1
Title:

A Comparative Study of Visualizations for Multiple Time Series
Authors:

Max Franke, Moritz Knabben, Julian Lang, Steffen Koch and Tanja Blascheck
Abstract: Different visualization techniques are suited for visualizing data of multiple time series. Choosing an appropriate visualization technique depends on data characteristics and tasks. Previous work has explored such combinations of data and visualization techniques in lab-based studies to find the most suited technique for a task. Using these previous findings, we performed an online study with 51 participants, during which we compare line charts, stream graphs, and aligned area charts based on completion time and accuracy regarding three common discrimination tasks. Our online study includes a novel combination of visualization techniques for time-dependent data and indicates that there are certain differences and trends regarding the suitability of the visualizations for different tasks. At the same time, we can confirm results presented in previous work.
Download
Paper Nr: 6
Title:

Lifting the Fog on Word Clouds: An Evaluation of Interpretability in 234 Individuals
Authors:

M. P. Maurits, M. Boers and R. Knevel
Abstract: Word clouds are a popular tool for text summary visualisation. By scaling words based on relative frequency, readers should be capable of quickly deducing some of the text semantics. We raise the question whether word clouds truly aid visualisation or rather mislead readers by scaling the wrong text aspect. We evaluated the magnitude of misinterpretation of word clouds using both a traditional font-scaling approach and a novel surface-area-scaling approach. Using an online survey we involved 234 participants, whom we tasked with guesstimating the frequency of 2 words either side of a word with a fixed frequency. We defined an error margin based on the regression slope of the guesstimations with the true frequencies. Clouds were constructed using the font-size or the word-area scaling method, a doubling or a linearly increasing frequency scheme and either words with a constant or increasing length. Errors were compared between settings using Wilcoxon tests. Both word size scaling methods resulted in poor performance of the participants and highlighted great inter-participant variation. Guesstimation accuracy was clearly dependent on the objective complexity of the visualisation. Our survey supports the hypothesis that word clouds are a fickle measure to convey word frequencies in a corpus of text.
Download
Paper Nr: 22
Title:

Panoptic Visual Analytics of Eye Tracking Data
Authors:

Valeria Garro and Veronica Sundstedt
Abstract: In eye tracking data visualization, areas of interest (AOIs) are widely adopted to analyze specific regions of the stimulus. We propose a visual analytics tool that leverages panoptic segmentation to automatically divide the whole image or frame video in semantic AOIs. A set of AOI-based visualization techniques are available to analyze the fixation data based on these semantic AOIs. Moreover, we propose a modified version of radial transition graph visualizations adapted to the extracted semantic AOIs and a new visualization technique also based on radial transition graphs. Two application examples illustrate the potential of this approach and are used to discuss its usefulness and limitations.
Download
Paper Nr: 9
Title:

Designing Animated Transitions for Dynamic Streaming Big Data
Authors:

João Moreira, Filipa Castanheira, Daniel Mendes and Daniel Gonçalves
Abstract: Visualizations for Streaming Big Data need to handle high volumes of information in real-time, making it challenging to convey significant data changes without confusing users. A simple first approach would be switching from the current visual idiom to another, highlighting a significant change. Unfortunately, there are no guidelines to design effective transitions between two visual idioms in Streaming Big Data. Therefore, we created a tree of animation concepts to serve as a starting point for designing such animated transitions. The concepts represent several ways in which a visual idiom can be transformed into another. We chose three visual idioms to test our idea and arranged several concepts to apply at each possible pairing (six possibilities). For each pairing, we tested the accuracy of people’s perceptions. Finally, we conducted a user study with 100 participants, where each participant answered various questions about transitions between two visual idioms shown in several videos. We concluded that to conceive appropriate animated transitions for Streaming Big Data (which also applies just for Data Streaming) that allow users to understand the changes in incoming data, varying how the proposed concepts are applied is not enough, highlighting the need for future research to address this challenge.
Download
Paper Nr: 12
Title:

Visualization of Activity Data from a Sensor-based Long-term Monitoring Study at a Playground
Authors:

Tobias Bolten, Regina Pohle-Fröhlich, Dorothee Volker, Clemens Brück, Nicolas Beucker and Hans-Günter Hirsch
Abstract: In the context of urban planning, a detailed knowledge of the considered space and its utilization is essential. However, manual observations are often not performed due to cost. Whereas sensor-based systems are often not installed due to possible constrains caused by data protection laws and user privacy-concerns. We addressed these concerns and developed a privacy-aware, sensor-based processing pipeline for detecting objects based on an analysis of signals from several sensors. These detections are used for their mapping and visualization in a global bird eye view. Besides a data normalization, which is crucial considering sections of different lengths, multiple variations of activity visualization applying heat maps are described. This includes the utilization of background representations with different levels of details, different accumulations of object detections through the adjustment of the performed spatial binning as well as applying different colormaps. Both sequential colormaps and diverging colormaps with and without perceptually uniform distances were considered. These variations were evaluated in a conducted online survey addressing professional urban planners as well as interested citizens. The results of this survey were used to determine a meaningful default setup for visualizing the activities in an interactive graphical user interface. This interface is intended to make the results of the performed long-term monitoring generally accessible.
Download
Paper Nr: 19
Title:

Feature-level Approach for the Evaluation of Text Classification Models
Authors:

Vanessa Bracamonte, Seira Hidano, Toru Nakamura and Shinsaku Kiyomoto
Abstract: Visualization of explanations of text classification models is important for their evaluation. The evaluation of these models is mostly based on visualization techniques that apply to a datapoint level. Although a feature-level evaluation is possible with current visualization libraries, existing approaches do not yet implement ways for an evaluator to visualize how a text classification model behaves for features of interest for the whole data or a subset of it. In this paper, we describe and evaluate a simple feature-level approach that leverages existing interpretability methods and visualization techniques to provide evaluators information on the importance of specific features in the behavior of a text classification model. We conduct case studies of two types of text classification models: a movie review sentiment classification model and a comment toxicity model. The results show that a feature-level explanation visualization approach can help identify problems with the models.
Download
Paper Nr: 29
Title:

Comparative Assessment of Two Data Visualizations to Communicate Medical Test Results Online
Authors:

Federico Cabitza, Andrea Campagner and Enrico Conte
Abstract: As most countries in the world still struggle to contain the COVID-19 breakout, Data Visualization tools have become increasingly important to support decision-making under uncertain conditions. One of the challenges posed by the pandemic is the early diagnosis of COVID-19: To this end, machine learning models capable of detecting COVID-19 on the basis of hematological values have been developed and validated. This study aims to evaluate the potential of two Data Visualizations to effectively present the output of a COVID-19 diagnostic model to render it online. Specifically, we investigated whether any visualization is better than the other in communicating a COVID-19 test results in an effective and clear manner, both with respect to positivity and to the reliability of the test itself. The findings suggest that designing a visual tool for the general public in this application domain can be extremely challenging for the need to render a wide array of outcomes that can be affected by varying levels of uncertainty.
Download

Area 3 - Spatial Data Visualization

Full Papers
Paper Nr: 26
Title:

ANTENNA: A Tool for Visual Analysis of Urban Mobility based on Cell Phone Data
Authors:

Pedro Silva, Catarina Maçãs, João Correia, Penousal Machado and Evgheni Polisciuc
Abstract: Nowadays, the collection of data from ubiquitous urban sensors, such as smartphones, can be used to analyse, understand, and profile urban mobility. This analysis requires dynamic, autonomous, and effective ways to parse, reduce and retrieve mobility patterns from large heterogeneous datasets. In this design study, we present ANTENNA, a visual analysis tool that allows the identification and analysis of urban mobility patterns based on mobile cell phone data. In particular, we present a visualization that is prepared for multiple scenarios of analysis, providing specific visualization approaches for different sets of tasks. We developed diverse visualization models to characterise inter- and intra-urban mobility. To validate ANTENNA, we conducted user tests with experts of different domains. The results suggest the appropriateness and usefulness of ANTENNA for each of the presented usage scenarios.
Download
Short Papers
Paper Nr: 5
Title:

Geo-Referenced Occlusion Models for Mixed Reality Applications using the Microsoft HoloLens
Authors:

Christoph Praschl and Oliver Krauss
Abstract: Emergency responders or task forces can benefit from outdoor Mixed Reality (MR) trainings, as they allow more realistic and affordable simulations of real-world emergencies. Utilizing MR devices for outdoor situations requires knowledge of real-world objects in the training area, enabling the realistic immersion of both, the real, as well as the virtual world, based on visual occlusions. Due to spatial limitations of state-of-the-art MR devices recognizing distant real-world items, we present an approach for sharing geo-referenced 3D geometries across multiple devices utilizing the CityJSON format for occlusion purposes in the context of geospatial MR visualization. Our results show that the presented methodology allows accurate conversion of occlusion models to geo-referenced representations based on a quantitative evaluation with an average error according to the vertices’ position from 1.30E-06 to 2.79E-04 (sub-millimeter error) using a normalized sum of squared errors metric. In the future, we plan to also incorporate 3D reconstructions from smartphones and drones to increase the number of supported devices for creating geo-referenced occlusion models.
Download