Overcome Selective Exposure in Web Search by considering Eye Movements and Physiological Signals

• Applicants: Dr. Daniel Hienert; Dr. Dagmar Kern
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 525041402

The amount of information on the web is ever-increasing, so people develop strategies to select important news and information and avoid cognitive load. Thereby, people prefer selecting and reading information that is consistent with their own. This phenomenon is explained by the psychological concepts of “selective exposure” and “confirmation bias” which can negatively affect personal decision-making and might lead to misjudgments. In this project, we will combine knowledge from the research areas of interactive information retrieval, knowledge discovery, and human-computer interaction to investigate options for counteracting selective exposure on the web. The overall goal is to support users in overcoming selective exposure behavior and make suggestions to inform themselves holistically about a topic. Our goals are the following: (1) Investigating how selective exposure manifests itself in interaction, behavioral, eye-tracking, and physiological data. (2) Based on this data, examining machine learning models for real-time detection of selective exposure. (3) Exploring how user interface components should be designed to make the users aware of selective exposure and motivate them to consume news with different thematic aspects of the overall topic. To this end, we first collect interaction, behavioral, eye-tracking, and physiological data in two lab studies. In the first study, we look at selective exposure during exploratory web search in a real-world setting. In the second study, we focus primarily on the aspect of reading news in a controlled environment to collect further physiological data accompanying the reading of articles that either confirm or contradict one's opinion. Second, we build a software pipeline to detect selective exposure based on behavioral data automatically. This includes a component that automatically identifies which aspects of an overall topic have been consumed by participants and which have not. Third, we want to provide meaningful awareness components that make selective exposure perceptible and point users to supplemental aspects. Therefore, we will evaluate different interaction designs with users. In the final step, we will combine all insights and developed components and conduct a user study in which selective exposure behavior in real-world web sessions is automatically detected and the participant is made aware.

• DFG Programme: Research Grants