Privacy

Hark automates the entire process of summarizing privacy feedback, starting from unstructured text and resulting in a hierarchy of high-level privacy themes and fine-grained issues within each theme, along with representative reviews for each issue.

Using CamForensics, we characterize how over 600 Android apps extract information such as text, faces, and QR codes from devices’ camera. In addition, we perform several surveys to characterize what information users expected these apps to extract based on their app store descriptions and to gauge their attitudes toward visual privacy. Our results show that apps frequently defy users’ expectations, based on their descriptions, and that users care about how apps process their camera data.

In this work, we present experimental results that utilize our campus Software-Defined Networking infrastructure to characterize how network-path latency, packet loss, and geographic distance impact offloading to the edge.

This paper made the case that privacy markers are a promising way to prevent third-party apps from inadvertently leaking visual information. We designed and implemented two privacy-marker systems that help users mark public regions in a camera’s view and deliver only content within the public regions to apps.

A Deep Learning System for Navigating Privacy Feedback at Scale