Paper Reading #17 : Privacy Risks Emerging from the Adoption of Innocuous Wearable Sensors in the Mobile Environment

Reference

Authors and Affiliations

Andrew Raij,University of South Florida

Animikh Ghosh,SETLabs, Infosys

Santosh Kumar,University of Memphis

Mani Srivastava, University of California Los Angeles

Presentation: CHI 2011, May 7–12, 2011, Vancouver, BC, Canada.

Summary

Hypothesis

The paper does a study to assess how concerned people are about disclosure of a variety of behaviors and contexts that are embedded in wearable sensor data. They hypothesize that these concerns are mediated by temporal and physical context associated with the data and the participant’s personal stake in the data.

Content

The three different goals for the project were were 1)Assess the privacy concerns of real people regarding disclosure of continuously collected physiological, behavioral, and psychological data 2)Use the proposed privacy framework and examine how restrictions and abstractions applied to various behaviors and contexts change concern levels. 3)Assess how reidentification of the data producer affects concern levels as the type of data consumer is varied. The different context used were temporal, physical, social and psychological.

Methods

User studies were done among 66 participants taking into consideration three different goals among two different groups of people ;a group with no personal stake in the data (Group NS) and a group with a personal stake (Group S) . For three days, Group S participants collected physiological, behavioral, and psychological data using the AutoSense sensor system as they went about their normal everyday life. Later they completed a privacy questionnaire assessing their concern regarding disclosure of selected behaviors and contexts with various restrictions and abstractions applied. Then the graphs were studied and a final questionnaire was asked to be filled. The other group only completed the same privacy questionnaire as Group S. The authors developed the Aha visualization system for reviewing data collected by the AutoSense system from natural environments.

Results

Group NS and Group S-Pre provided similar concern ratings for exercise and place, and no significant differences were found between their distributions. However, there was a trend toward significance for commuting .Overall results indicate increasing personal stake in the data helps people better estimate their concerns regarding the disclosure of behaviors and contexts. The highest level of concerns emerged for exercise, conversation, commuting, and stress after participants observed visual depictions of their data.

Adding temporal and physical context to behaviors and other contexts increased concern level, with each decreasing level of temporal abstraction leading to a corresponding increase in level of concern. Adding timestamps induced a significant change in concern as compared to adding duration.Adding both place and time simultaneously rather than alone increased concerns significantly for conversation, commuting, and stress.

Participants were more concerned on releasing identified data to other researchers to releasing identified data to the general public.The results indicate disclosure of any data to the general pub- lic is of significant concern to participants. When identity is added to the dataset, this concern more than doubles.

Discussion

The research and results were really impressive to me. There are many cases when we are providing more information about ourselves than needed and this paper does a good job in pointing out that issue in the field of wearable sensors. The research should make the future innovations and technologies to take into consideration the privacy of user as well as make people aware that they might be providing unintended information and never realize that.