Information Studies

Permanent URI for this communityhttp://hdl.handle.net/1903/2249

Browse

Subject: search.filters.subject.machine learning

Search Results

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    Perceptual Pat: A Virtual Human Visual System for Iterative Visualization Design
    (Association for Computer Machinery (ACM), 2023-04-23) Shin, Sungbok; Hong, Sanghyun; Elmqvist, Niklas
    Designing a visualization is often a process of iterative refinement where the designer improves a chart over time by adding features, improving encodings, and fixing mistakes. However, effective design requires external critique and evaluation. Unfortunately, such critique is not always available on short notice and evaluation can be costly. To address this need, we present Perceptual Pat, an extensible suite of AI and computer vision techniques that forms a virtual human visual system for supporting iterative visualization design. The system analyzes snapshots of a visualization using an extensible set of filters—including gaze maps, text recognition, color analysis, etc—and generates a report summarizing the findings. The web-based Pat Design Lab provides a version tracking system that enables the designer to track improvements over time. We validate Perceptual Pat using a longitudinal qualitative study involving 4 professional visualization designers that used the tool over a few days to design a new visualization.
  • Thumbnail Image
    Item
    USING SOCIAL MEDIA AS A DATA SOURCE IN PUBLIC HEALTH RESEARCH
    (2022) Sigalo, Nekabari; Frias-Martinez, Vanessa; Library & Information Services; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Researchers have increasingly looked to social media data as a means of measuring population health and well-being in a less intrusive and more scalable manner compared to traditional public health data sources. In this dissertation, I outline three studies that leverage social media as a data source, to answer research questions related to public health and compare traditional public health data sources to social media data sources. In Study #1, I conduct a study with the aim of developing, from geotagged Twitter data, a predictive model for the identification of food deserts in the United States, using the linguistic constructs found in food-related tweets. The results from this study suggest the food-ingestion language found in tweets, such as census-tract level measures of food sentiment and healthiness, are associated with census tract-level food desert status. Additionally, the results suggest that including food ingestion language derived from tweets in classification models that predict food desert status improves model performance when compared to baseline models that only include socio-economic characteristics. In Study #2, I evaluate whether attitudes towards COVID-19 vaccines collected from the Household Pulse Survey can be predicted using attitudes extracted from Twitter. The results reveal that attitudes toward COVID-19 vaccines found in tweets explain 61-72% of the variability in the percentage of HPS respondents that were vaccine hesitant or compliant. The results also reveal significant statistical relationships between perceptions expressed on Twitter and in the survey. In Study #3, I conduct a study to examine whether supplementing COVID-19 vaccine uptake forecast models with the attitudes found in tweets improves over baseline models that only use historical vaccination data. The results of this study reveal that supplementing baseline forecast models with both historical vaccination data and COVID-19 vaccine attitudes found in tweets reduce RMSE by as much as 9%. The studies outlined in this dissertation suggest there is a valuable signal for public health research in Twitter data.
  • Thumbnail Image
    Item
    Understanding and Intervening in Machine Learning Ethics: Supporting Ethical Sensitivity in Training Data Curation
    (2020) Boyd, Karen L; Shilton, Katie; Library & Information Services; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Despite a great deal of attention to developing mitigations for ethical concerns in Machine Learning (ML) training data and models, we don’t yet know how these interventions will be adopted and used. Will they help ML engineers find and address ethical concerns in their work? This dissertation seeks to understand ML engineers’ ethical sensitivity (ES)— their propensity to notice, analyze, and act on socially impactful aspects of their work—while curating training data. A systematic review of ES (Chapter 2) addresses conflicts of conceptualization in prior work by developing a new framework describing three activities (recognition, particularization, and judgment); argues that ES offers a useful way to describe, evaluate, and intervene in ethical technology development; and argues that the methods and perspectives of social computing can offer richer methods and data to studies of ES. A think aloud study (Chapter 3) tests this framework by using ES to compare engineers working with unfamiliar training data, finding that engineers with Datasheets noticed ethical issues earlier and more frequently than those without; finding that participants relied on Datasheets extensively while particularizing; and rendering rich descriptions of recognition and particularization in facial recognition data curation. Chapter 4 uses Value Sensitive Design to "design up,'' mitigating harms by helping machine learning engineers particularize their ethical concerns and find appropriate technical tools. It introduces ES to studies of social computing, contributes a novel method for studying ES, offers rich data about how it functions in ML development, describes insights for designing context documents and other interventions designed to encourage ES, develops an extensible digital guide that supports particularization and judgment, and points to new directions for research in ethical sensitivity in technology development.
  • Thumbnail Image
    Item
    Local Information Landscapes: Theory, Measures, and Evidence
    (2019) Lee, Myeong; Butler, Brian S; Geography/Library & Information Systems; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    To understand issues about information accessibility within communities, research studies have examined human, social, and technical factors by taking a socio-technical view. While this view provides a profound understanding of how people seek, use, and access information, this approach tends to overlook the impact of the larger structures of information landscapes that constantly shape people’s access to information. When it comes to local community settings where local information is embedded in diverse material entities such as urban places and technical infrastructures, the effect of information landscapes should be taken into account in addition to particular strategies for solving information-seeking issues. However, characterizing the information landscape of a local community at the community level is a non-trivial problem due to diverse contexts, users, and their interactions with each other. One way to conceptualize local information landscapes in a way that copes with the complexity of the interplay between information, contexts, and human factors is to focus on the materiality of information. By focusing on the material aspects of information, it becomes possible to understand how local information is provided to social entities and infrastructures and how it exists, forming structures at the community level. Through an extensive literature review, this paper develops a theory of local information landscapes (LIL Theory) to better conceptualize the community-level, material structure of local information. Specifically, the LIL theory adapts a concept of the virtual as an ontological view of the interplay between technical infrastructures, spaces, and people as a basis for assessing and explaining community-level structures of local information. By complementing existing theories such as information worlds and information grounds, this work provides a new perspective on how information deserts manifest as a material pre-condition of information inequality. Using this framework, an empirical study was conducted to examine the explicit effects of information deserts on other community characteristics. Specifically, the study aims to provide an initial assessment of LIL theory by examining how the fragmentation of local information, a form of information deserts, is related to important community characteristics such as socio-economic inequality, deprivation, and community engagement. Building upon previous work in sociology and political science, this study shows that the fragmentation of local information (1) is shaped by socio-economic deprivation/inequality that is confounded with ethnoracial heterogeneity, (2) the fragmentation of local information is highly correlated to people's community gatherings, (3) the fragmentation of local information moderates the effects of socio-economic inequality on cultural activity diversity, and (4) the fragmentation of local information mediates the relationship between socio-economic inequality and community engagement. By making use of three local event datasets over 20 months in 14 U.S. cities (about two million records) and over 3 months in 28 U.S. cities (about 620K records), respectively, this study develops computational frameworks to operationalize information deserts in a scalable way. This dissertation provides a theorization of community-level information inequality and computational models that support the quantitative examination of it. Further theorizations of the conceptual constructs and methodological improvements on measurements will benefit information policy-makers, local information system designers, and researchers who study local communities with conceptual models, vocabularies, and assessment frameworks.