Learning Privacy Preferences from Real-World Decisions

Aufgabenstellung / Topic

Users regularly face privacy-related decisions while browsing the web, such as responding to cookie banners, granting permissions, or configuring tracking settings. Making these decisions consistently across different websites and contexts is challenging, and many users experience decision fatigue or make choices that do not match their actual preferences. Recent research suggests that automated privacy assistants could observe user behavior and recommend privacy decisions, potentially reducing cognitive load and improving alignment with user preferences. However, it remains unclear how accurately user preferences can be predicted in real-world situations, how users respond to such recommendations, and how much control they wish to retain. In this thesis, you will develop a browser extension that observes real privacy decisions, generates simple recommendations based on past behavior, and allows users to confirm or override the suggestions. The goal is to investigate how predictable privacy decisions are, how users perceive automated recommendations, and how interactions with the assistant influence trust and behavior.

You will:

• Review literature on privacy decision-making, preference modeling, and automated privacy assistants
• Develop a browser extension that tracks privacy decisions (e.g., cookie consent, tracker blocking, permissions)
• Implement a simple model that predicts user decisions based on prior behavior
• Conduct a user study to evaluate prediction accuracy, user responses, and trust in the assistant
• Analyze whether automated suggestions align with user preferences and identify patterns in overrides
• Summarize findings in a thesis and (optionally) contribute to a research publication

You need:

• Interest in usable privacy, human-computer interaction, and AI-assisted systems
• Programming skills (JavaScript for browser extensions, Python/R for data analysis)
• Basic knowledge of machine learning or preference modeling
• Motivation to conduct user studies and analyze behavioral data
• Attention to ethical considerations in observing real user behavior