Publication Details
| Florian Bemmann
Privacy Preserving User Quantification with Smartphone Sensing Upcoming (bib) |
|
| With smartphones becoming omnipresent in peopleâs everyday lives, mobile sensing data logging and usage have proliferated widely in the last decade. However, perceived privacy concerns and fears raise skepticism among smartphone users and finally throttle the spread and use of mobile sensing-based apps. To accommodate security and privacy issues, operating system developers impose general access restrictions on some kinds of data to protect the users. This obstructs the development of novel, adaptive intelligent interfaces and the study of better privacy-enhancing technologies. In this thesis, we study how we can improve the privacy of smartphone users while keeping the data usable for application purposes. We first motivate the demand for mobile sensing data by showing what information can be extracted with state-of-theart mobile sensing systems, which use cases they fuel, and how three stakeholders, namely users, researchers, and society, could benefit from their implementation. After that, we highlight the emerging privacy issues and show how much they hinder app adoption. Finally, we propose user-centered approaches that mitigate these privacy issues while keeping the output that fuels the dataâs use cases. We found an overall lack of user-centeredness and identified transparency and control features as key elements towards an improved privacy perception. While offering control features directly reduces privacy concerns, transparency initially worsens the situation unless it is applied comprehensively. We outline how transparency and control can be integrated with beneficial effects, discuss different approaches to contextual privacy, the (in)appropriateness of privacy interfaces in situ, and the issue of a lack of user motivation regarding privacy belongings. My thesis contributes to a better understanding of privacy in smartphone sensing. It facilitates using mobile sensing data for adaptive intelligent applications while preserving the usersâ privacy. |
