Browsing by Author "Dunne, Cody"

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Ask and You Shall Receive (a Graph Drawing): Testing ChatGPT's Potential to Apply Graph Layout Algorithms
    (The Eurographics Association, 2023) Bartolomeo, Sara Di; Severi, Giorgio; Schetinger, Victor; Dunne, Cody; Hoellt, Thomas; Aigner, Wolfgang; Wang, Bei
    Large language models (LLMs) have recently taken the world by storm. They can generate coherent text, hold meaningful conversations, and be taught concepts and basic sets of instructions-such as the steps of an algorithm. In this context, we are interested in exploring the application of LLMs to graph drawing algorithms by performing experiments on ChatGPT, one of the most recent cutting-edge LLMs made available to the public. These algorithms are used to create readable graph visualizations. The probabilistic nature of LLMs presents challenges to implementing algorithms correctly, but we believe that LLMs' ability to learn from vast amounts of data and apply complex operations may lead to interesting graph drawing results. For example, we could enable users with limited coding backgrounds to use simple natural language to create effective graph visualizations. Natural language specification would make data visualization more accessible and user-friendly for a wider range of users. Exploring LLMs' capabilities for graph drawing can also help us better understand how to formulate complex algorithms for LLMs; a type of knowledge that could transfer to other areas of computer science. Overall, our goal is to shed light on the exciting possibilities of using LLMs for graph drawing-using the Sugiyama algorithm as a sample case-while providing a balanced assessment of the challenges and opportunities they present. A free copy of this paper with all supplemental materials to reproduce our results is available on osf.io .
  • Thumbnail Image
    Item
    Daisen: A Framework for Visualizing Detailed GPU Execution
    (The Eurographics Association and John Wiley & Sons Ltd., 2021) Sun, Yifan; Zhang, Yixuan; Mosallaei, Ali; Shah, Michael D.; Dunne, Cody; Kaeli, David; Borgo, Rita and Marai, G. Elisabeta and Landesberger, Tatiana von
    Graphics Processing Units (GPUs) have been widely used to accelerate artificial intelligence, physics simulation, medical imaging, and information visualization applications. To improve GPU performance, GPU hardware designers need to identify performance issues by inspecting a huge amount of simulator-generated traces. Visualizing the execution traces can reduce the cognitive burden of users and facilitate making sense of behaviors of GPU hardware components. In this paper, we first formalize the process of GPU performance analysis and characterize the design requirements of visualizing execution traces based on a survey study and interviews with GPU hardware designers. We contribute data and task abstraction for GPU performance analysis. Based on our task analysis, we propose Daisen, a framework that supports data collection from GPU simulators and provides visualization of the simulator-generated GPU execution traces. Daisen features a data abstraction and trace format that can record simulator-generated GPU execution traces. Daisen also includes a web-based visualization tool that helps GPU hardware designers examine GPU execution traces, identify performance bottlenecks, and verify performance improvement. Our qualitative evaluation with GPU hardware designers demonstrates that the design of Daisen reflects the typical workflow of GPU hardware designers. Using Daisen, participants were able to effectively identify potential performance bottlenecks and opportunities for performance improvement. The open-sourced implementation of Daisen can be found at gitlab.com/akita/vis. Supplemental materials including a demo video, survey questions, evaluation study guide, and post-study evaluation survey are available at osf.io/j5ghq.
  • Thumbnail Image
    Item
    Effective Use of Likert Scales in Visualization Evaluations: A Systematic Review
    (The Eurographics Association and John Wiley & Sons Ltd., 2022) South, Laura; Saffo, David; Vitek, Olga; Dunne, Cody; Borkin, Michelle A.; Borgo, Rita; Marai, G. Elisabeta; Schreck, Tobias
    Likert scales are often used in visualization evaluations to produce quantitative estimates of subjective attributes, such as ease of use or aesthetic appeal. However, the methods used to collect, analyze, and visualize data collected with Likert scales are inconsistent among evaluations in visualization papers. In this paper, we examine the use of Likert scales as a tool for measuring subjective response in a systematic review of 134 visualization evaluations published between 2009 and 2019. We find that papers with both objective and subjective measures do not hold the same reporting and analysis standards for both aspects of their evaluation, producing less rigorous work for the subjective qualities measured by Likert scales. Additionally, we demonstrate that many papers are inconsistent in their interpretations of Likert data as discrete or continuous and may even sacrifice statistical power by applying nonparametric tests unnecessarily. Finally, we identify instances where key details about Likert item construction with the potential to bias participant responses are omitted from evaluation methodology reporting, inhibiting the feasibility and reliability of future replication studies. We summarize recommendations from other fields for best practices with Likert data in visualization evaluations, based on the results of our survey.
  • Thumbnail Image
    Item
    Process and Pitfalls of Online Teaching and Learning with Design Study ''Lite'' Methodology: A Retrospective Analysis
    (The Eurographics Association and John Wiley & Sons Ltd., 2023) Syeda, Uzma Haque; Dunne, Cody; Borkin, Michelle A.; Bujack, Roxana; Archambault, Daniel; Schreck, Tobias
    Design studies are an integral method of visualization research with hundreds of instances in the literature. Although taught as a theory, the practical implementation of design studies is often excluded from visualization pedagogy due to the lengthy time commitments associated with such studies. Recent research has addressed this challenge and developed an expedited design study framework, the Design Study ''Lite'' Methodology (DSLM), which can implement design studies with novice students within just 14 weeks. The framework was developed and evaluated based on five semesters of in-person data visualization courses with 30 students or less and was implemented in conjunction with Service-Learning (S-L).With the growth and popularity of the data visualization field-and the teaching environment created by the COVID-19 pandemic-more academic institutions are offering visualization courses online. Therefore, in this paper, we strengthen and validate the epistemological foundations of the DSLM framework by testing its (1) adaptability to online learning environments and conditions and (2) scalability to larger classes with up to 57 students. We present two online implementations of the DSLM framework, with and without Service-Learning (S-L), to test the adaptability and scalability of the framework. We further demonstrate that the framework can be applied effectively without the S-L component. We reflect on our experience with the online DSLM implementations and contribute a detailed retrospective analysis using thematic analysis and grounded theory methods to draw valuable recommendations and guidelines for future applications of the framework. This work verifies that DSLM can be used successfully in online classes to teach design study methodology. Finally, we contribute novel additions to the DSLM framework to further enhance it for teaching and learning design studies in the classroom.
  • Thumbnail Image
    Item
    Six Methods for Transforming Layered Hypergraphs to Apply Layered Graph Layout Algorithms
    (The Eurographics Association and John Wiley & Sons Ltd., 2022) Bartolomeo, Sara Di; Pister, Alexis; Buono, Paolo; Plaisant, Catherine; Dunne, Cody; Fekete, Jean-Daniel; Borgo, Rita; Marai, G. Elisabeta; Schreck, Tobias
    Hypergraphs are a generalization of graphs in which edges (hyperedges) can connect more than two vertices-as opposed to ordinary graphs where edges involve only two vertices. Hypergraphs are a fairly common data structure but there is little consensus on how to visualize them. To optimize a hypergraph drawing for readability, we need a layout algorithm. Common graph layout algorithms only consider ordinary graphs and do not take hyperedges into account. We focus on layered hypergraphs, a particular class of hypergraphs that, like layered graphs, assigns every vertex to a layer, and the vertices in a layer are drawn aligned on a linear axis with the axes arranged in parallel. In this paper, we propose a general method to apply layered graph layout algorithms to layered hypergraphs. We introduce six different transformations for layered hypergraphs. The choice of transformation affects the subsequent graph layout algorithm in terms of computational performance and readability of the results. Thus, we perform a comparative evaluation of these transformations in terms of number of crossings, edge length, and impact on performance. We also provide two case studies showing how our transformations can be applied to real-life use cases.