Computer Science Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2756
Browse
7 results
Search Results
Item SIMULATION, REPRESENTATION, AND AUTOMATION: HUMAN-CENTERED ARTIFICIAL INTELLIGENCE FOR AUGMENTING VISUALIZATION DESIGN(2024) Shin, Sungbok; Elmqvist, Niklas; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Data visualization is a powerful strategy for using graphics to represent data for effective communication and analysis. Unfortunately, creating effective data visualizations is a challenge for both novice and expert design users. The task often involves an iterative process of trial and error, which by its nature, is time-consuming. Designers frequently seek feedback to ensure their visualizations convey the intended message clearly to their target audience. However, obtaining feedback from peers can be challenging, and alternatives like user studies or crowdsourcing is costly and time-consuming. This suggests the potential for a tool that can provide design feedback for visualizations. To that end, I create a virtual, human vision-inspired system that looks into the visualization design and provides feedback on it using various AI techniques. The goal is not to replicate an exact version of a human eye. Instead, my work aims to develop a practical and effective system that delivers design feedback to visualization designers, utilizing advanced AI techniques, such as deep neural networks (DNNs) and large language models (LLMs). My thesis includes three distinct works, each aimed at developing a virtual system inspired by human vision using AI techniques. Specifically, these works focus on simulation, representation, and automation, collectively progressing toward the aim. First, I develop a methodology to simulate human perception in machines through a virtual eye tracker named A SCANNER DEEPLY. This involves gathering eye gaze data from chart images and training them using a DNN. Second, I focus on effectively and pragmatically representing a virtual human vision-inspired system by creating PERCEPTUAL PAT, which includes a suite of perceptually-based filters. Third, I automate the feedback generation process with VISUALIZATIONARY, leveraging large language models to enhance the automation. I report on challenges and lessons learned about the key components and design considerations that help visualization designers. Finally, I end the dissertation by discussing future research directions for using AI for augmenting visualization design process.Item Information Olfactation: Theory, Design, and Evaluation(2019) Patnaik, Biswaksen; Elmqvist, Niklas; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Olfactory feedback for analytical tasks is a virtually unexplored area in spite of the advantages it offers for information recall, feature identification, and location detection. Here we introduce the concept of ‘Information Olfactation’ as the fragrant sibling of information visualization, and discuss how scent can be used to convey data. Building on a review of the human olfactory system and mirroring common visualization practice, we propose olfactory marks, the substrate in which they exist, and their olfactory channels that are available to designers. To exemplify this idea, we present ‘viScent(1.0)’: a six-scent stereo olfactory display capable of conveying olfactory glyphs of varying temperature and direction, as well as a corresponding software system that integrates the display with a traditional visualization display. We also conduct a comprehensive perceptual experiment on Information Olfactation: the use of olfactory marks and channels to convey data. More specifically, following the example from graphical perception studies, we design an experiment that studies the perceptual accuracy of four ``olfactory channels''---scent type, scent intensity, airflow, and temperature---for conveying three different types of data---nominal, ordinal, and quantitative. We also present details of an advanced 24-scent olfactory display: ‘viScent(2.0)’ and its software framework that we designed in order to run this experiment. Our results yield a ranking of olfactory channels for each data type that follows similar principles as rankings for visual channels, such as those derived by Mackinlay, Cleveland & McGill, and Bertin.Item Designing Interactive Decision Aids for Medical Risk Communication and Exploration of Treatment Options(2013) Franklin, Lyndsey; Shneiderman, Ben; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Medical treatments carry unique benefits and risks which patients must understand in order to decide which of their options is best for them. Prior research has demonstrated that patients are ill-equipped to understand both medical terminology and the statistical information presented to them through standard decision aids. Patients are unable to use the information about treatments to make decisions and as a result make poor choices with regards to their healthcare. The contributions of this work are 1) a multi-dimensional model for describing the content of decision aids; 2) TreatmentExplorer, a prototype interactive decision aid designed to communicate treatment risks and benefits through the use of visualization, animation, and guided narration; 3) an evaluation of TreatmentExplorer with four experts in health communication; 4) a preliminary usability evaluation comparing the performance of TreatmentExplorer against design alternatives, and 5) guidelines for interactive decision aids based on the results of these preliminary user evaluations.Item E XPLORING DIFFERENCES IN MULTIVARIATE DATASETS USING HIERARCHIES AN INTERACTIVE INFORMATION VISUALIZATION APPROACH(2013) Guerra Gómez, John Alexis; Shneiderman, Ben; Plaisant, Catherine; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Hierarchies are a useful way of representing data. The parent-child relationships they define facilitate the analysis of a dataset by breaking it down into its component parts. Representing data as hierarchies can also be used to track changes to a dataset over time or between versions. For example, analysts can use hierarchies to uncover changes in the US Federal Budget in the last twenty years, by grouping accounts by Agencies and Bureaus. Similarly, a company manager can analyze changes to their product sales due to the holiday season by breaking them up by markets and product categories. Exploring differences in such trees could help them understand changes in the data. However, comparing hierarchies is a difficult task, even when comparing two trees with a small number of nodes. To address this, information visualization techniques were used to support data comparison tasks using hierarchies. After evaluating my techniques with domain experts on real world problems, I identified and addressed two main research topics: Abstract This dissertation first tackled the problem of comparing two versions of a tree by using two types of change, while most of the significant work on this topic has focused only on changes in node values or changes in topology. TreeVersity (http://hcil.cs.umd.edu/treeversity) is a comparison tool that allows users to explore changes between two versions of a tree by tracking node value differences, and newly created or removed nodes. Domain experts using TreeVersity were excited to discover differences in the trees, but expressed a desire to explore the evolution of a dataset over time. To that end, they suggested applying TreeVersity comparison capabilities to datasets that were non inherently hierarchical. Abstract Following users' feedback, the problem of exploring changes over time in datasets that can be categorized as trees was addressed next. TreeVersity2 (http://treeversity.cattlab.umd.edu is a web-based data comparison tool that allows users to explore a tree that changes over time and of datasets that are not inherently hierarchical, by categorizing them by their attributes. TreeVersity2 also helps users navigate the sometimes large amounts of differences between versions of a tree using an interactive textual reporting tool. Abstract My research has resulted in three main contributions: First, the introduction of the Bullet, a visualization glyph to represent four characteristics of change (as described in Section 1.2) in tree nodes, and the implementation of the Bullet in TreeVersity. Second, the creation of the StemView, a tree visualization technique that represents five characteristics of change in all the nodes of a tree (not just the leaves), and the implementation of the StemView in TreeVersity2. Furthermore, my research resulted in the development of the reporting tool, another feature of TreeVersity2, which helps users navigate outstanding changes in the tree with textual representations and coordinated interactions. Third, the development of 13 case studies with domain experts on real world comparison problems. The case studies have validated the utility and flexibility of my approaches. Finally, my research opens possibilities for future research on comparing hierarchical structures.Item Interactive Visual Displays for Results Management in Complex Medical Workflows(2013) Tarkan, Sureyya; Shneiderman, Ben A.; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Clinicians manage medical orders to ensure that the results are returned promptly to the correct physician and followed up on time. Delays in results management occur frequently, physically harm patients, and often cause malpractice litigation. Better tracking of medical orders that showed progress and indicated delays, could result in improved care, better safety, and reduced clinician effort. This dissertation presents novel displays of rich tables with an interaction technique called ARCs (Actions for Rapid Completion). Rich tables are generated by MStart (Multi-Step Task Analyzing, Reporting, and Tracking) from a workflow model that defines order processes. Rich tables help clinicians perceive each order's status, prioritize the critical ones, and act on results in a timely fashion. A second contribution is the design of an interactive visualization called MSProVis (Multi-Step Process Visualization), which is composed of several PCDs (Process Completion Diagrams) that show the number and duration of in-time, late, and not-completed orders. With MSProVis, managers perform retrospective analyses to make decisions by studying an overview of the order process, durations of order steps, and performances of individuals. I visited seven hospitals and clinics to define sample results management workflows. Iterative design reviews with clinicians, designers, and researchers led to refinements of the rich tables, ARCs, and design guidelines. A controlled experiment with 18 participants under time pressure and distractions tested two features (showing pending orders and prioritizing by lateness) of rich tables. These changes statistically significantly reduce the time from nine to one minute to correctly identify late orders compared to the traditional chronologically-ordered lists. Another study demonstrated that ARCs speed performance up by 25% compared to state-of-the-art systems. A usability study with two clinicians and five novices showed that participants were able to understand MSProVis and efficiently perform representative tasks. Two subjective preference surveys suggested new design choices for the PCDs. This dissertation provides designers of results management systems with clear guidance about showing pending results and prioritizing by lateness, and tested strategies for performing retrospective analyses. It also offers detailed design guidelines for results management, tables, and integrated actions on tables that speed performance for common tasks.Item Interactive Visualization Techniques for Searching Temporal Categorical Data(2010) Wang, Taowei David; Shneiderman, Ben; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Temporal data has always captured people's imagination. Large databases of temporal data contain temporal patterns that can lead to the discovery of important cause-and-effect phenomena. Since discovering these patterns is a difficult task, there is a great opportunity to improve support for searching. Temporal analysis of, for example, medical records, web server logs, legal, academic, or criminal records can benefit from more effective search strategies. This dissertation describes several interactive visualization techniques designed to enhance analysts' experience in performing search, exploration, and summarization of multiple sets of temporal categorical data. These techniques are implemented in the software Lifelines2 (http://www.cs.umd.edu/hcil/lifelines2/). Lifelines2 is an interactive visualization system that enables analysts to dynamically change their focus in order to expose temporal ordering of event sequences and study the prevalence of such orderings. This dissertation makes four main contributions. The first three are technical contributions, and the last is a process model that generalizes user behavior. First, the Align-Rank-Filter framework is presented to help analysts perform visual search and exploration. It enables analysts to center their attention on temporal events that are the focus of their inquiry. Through a controlled experiment, alignment alone is shown to improve user performance speed by up to 60% in tasks that require understanding of temporal ordering of events. The initial successful exploration on the alignment operator led to its fuller exploitation. Further enhancements to filtering are presented to better incorporate alignment. Second, I designed and implemented the Temporal Pattern Search (TPS) algorithm for filtering to support the common, but difficult-to-specify absence of operator in a temporal pattern. TPS exploits the data structure of the visualization system, and it compares favorably to existing common approaches. Third, I present the temporal summaries technique as an overview to support grouping and comparison features in Lifelines2. They support higher-level tasks such as hypothesis generation. These features take advantage of alignment, and the entirety of the system is evaluated in several long-term case studies with domain experts working on their own problems. Fourth, from these long-term case studies, I generalize a process model that describes analyst behavior in searching and interacting with temporal categorical data. Gleaning from observations in the case studies, collaborators' interviews and commentaries, and logs of Lifelines2 usage, I recommend feature design guidelines for future visualization designers for temporal categorical data. The enthusiasm of the domain experts who used Lifelines2, the changing strategies for problem-solving, and their initial successes suggest these interactive visualization techniques are a valuable addition to search capabilities.Item Interactive Visualizations for Trees and Graphs(2006-04-27) Lee, Bongshin; Bederson, Benjamin B; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Graphs are a very commonly used information structure, and have been applied to a broad range of fields from computer science to biology. There are several important issues to consider when designing graph visualizations. One of the most difficult problems researchers face is how to visualize large graphs. While an algorithm may produce good layouts for graphs of several hundred nodes, it may not scale well to several thousand nodes. And, as the size of the graph increases, performance will degrade rapidly, making it difficult to build an interactive system. Label readability will also suffer, hindering users' abilities to understand the graph data and perform many tasks. Finally, even if a system can lay out and display large graphs, the cognitive demands placed on the user by the visualization may be overwhelming. This dissertation describes and applies several design principles to various graph visualization domains to address these issues. Tightly-coupled and highly customized views were used for graph visualization in a novel way. A new tree layout approach to graph visualization was proposed with appropriate visualization and interaction techniques. When visualizing graphs as trees, a guiding metaphor "Plant a seed and watch it grow" was used to support information gathering and detailed exploration of the graph's local structure. Three graph visualization systems guided by these design principles were also developed and evaluated. First, PaperLens provides an abstract overview of the full dataset and shows relationships through interactive highlighting. It offers a novel alternative to the more common node-link diagram approach to graph visualization. Second, the development and evaluation of TaxonTree provided valuable insights that led to the design of TreePlus, a general interactive graph visualization component. Finally, TreePlus takes a tree layout approach to graph visualization, transforming a graph into a tree plus cross links (the links not represented by the spanning tree) using visualization, animation and interaction techniques to reveal the graph structure while preserving the label readability. Other contributions of this work include the development of a task taxonomy for graph visualization and several specific applications of the graph visualization systems described above.