2026-05-24T20:37:22Zhttps://keep.lib.asu.edu/oai/request

oai:keep.lib.asu.edu:node-1953592024-12-23T18:01:48Zoai_pmh:alloai_pmh:repo_items

195359 https://hdl.handle.net/2286/R.2.N.195359 http://rightsstatements.org/vocab/InC/1.0/ All Rights Reserved 2024 430 pages Doctoral Dissertation Academic theses Text eng Arunkumar, Anjana Bryan, Chris Maciejewski, Ross Baral, Chitta Bae, Gi-Yeul Arizona State University Partial requirement for: Ph.D., Arizona State University, 2024 Field of study: Computer Science Data visualization is essential for communicating complex information to diverse audiences. However, a gap persists between visualization design objectives and the understanding of non-expert users, with limited experience. This dissertation addresses challenges in designing for non-experts, referred to as the D.U.C.K. bridge: (i) user unfamiliarity with DATA analysis domains, (ii) variation in user UNDERSTANDING mechanisms, (iii) catering to individual differences in CREATING visualizations, and (iv) promoting KNOWLEDGE synthesis and application. By developing human-driven principles and tools, this research aims to enhance visualization creation and consumption by non-experts. Leveraging linked interactive visualizations, this dissertation explores the iterative education of non-experts when navigating unfamiliar DATA realms. VAIDA guides crowd workers in creating better NLP benchmarks through real-time visual feedback. Similarly, LeaderVis allows users to interactively customize AI leaderboards and select model configurations suited to their application. Both systems demonstrate how visual analytics can flatten the learning curve associated with complex data and technologies. Next, this dissertation examines how individuals internalize real-world visualizations—either as images or information. Experimental studies investigate the impact of design elements on perception across visualization types and styles, and an LSTM model predicts the framing of the recall process. The findings reveal mechanisms that shape the UNDERSTANDING of visualizations, enabling the design of tailored approaches to improve recall and comprehension among non-experts. This research also investigates how known design principles apply to CREATING visualizations for underrepresented populations. Findings reveal that multilingual individuals prefer varying text volumes based on annotation language, and older age groups engage more emotionally with affective visualizations than younger age groups. Additionally, underlying cognitive processes, like mind wandering, affect recall focus. These insights guide the development of more inclusive visualization solutions for diverse user demographics. This dissertation concludes by presenting projects aimed at preserving cognitive and affective KNOWLEDGE synthesized through visual analysis. The first project examines the impact of data visualizations in VR on personal viewpoints about climate change, offering insights for using VR in public scientific education. The second project introduces LINGO, which enables the creation of diverse natural language prompts for generative models across multiple languages, potentially facilitating custom visualization creation via streamlined prompting. Computer Science Computer Engineering Behavioral Sciences Behavioral Evaluation Computer Science Data Visualization decision making Natural Language Processing Visual Cognition The D.U.C.K. Bridge: Empowering Non-Experts in Data Visualization