Bachelorarbeit im Studiengang Informatik von Louis Pouliot: Chartchecker: A User-Centered Investigation of Chart Annotation and Misinformation Detection Software

Abstract: The ever increasing usage of social media as a news source allows for opinionated and deceptive content to be spread much more rapidly. An especially problematic aspect of this is the existence of misleading data visualizations. These graphs may, whether intentionally or not, violate design conventions, using them to alter the user’s perception of the data portrayed in them. This lets users draw false conclusions, even when the underlying data is factually correct. Together with the absence of effective methods for users to automatically identify misleading graphs this allows for the rampant spread of misinformation online, especially in times of public crises like the COVID-19 pandemic. We examined state of the art chart data extraction methods and used it to improve the existing chart annotation software “ChartChecker” in order to increase its usability and range of applicable charts. Subsequently, a user study was conducted involving 15 participants that got to utilize the tool in a monitored environment using real world charts. We gathered feedback regarding previous design choices as well as the functionality and comprehensibility of different aspects of the tool. Our findings shed light on users’ strategies when engaging with visualization annotation software. We also analyzed expectations that users have regarding the tools current features and interface. This allows us to create concrete guidelines for the future development of the ChartChecker tool as well as visualization misinformation annotation software in general.

Betreuer*in: Tom Biselli

Prüfer: Prof. Dr. Dr. Christian Reuter