Carleton CS Comps 2015-16: Privacy Manager
Privacy Manager
Advisor: Amy Csizmar Dalal
Background
Every day, we give away our personal data to countless web sites, apps, programs, etc. While on some level we may be aware that we are sharing personal information, we may not be aware of the extent to which we share this data, or exactly which data we share with which entities. Throw in things like cookies and analytics, and things muddy even further. (How many times have you noticed a Facebook ad related to that Google search you did a half hour ago?)
Ultimately, we choose to share personal information for a variety of reasons. Sharing our personal data with companies may get us perks, like coupons or free stuff or exclusive deals. Storing previous transaction information helps systems recommend things to us, preserves state across sessions, and makes authenticating easier and faster. But everything comes with a price, and we likely don't fully understand the extent to which our data is "out there" or the repercussions of our data promiscuity.
I envision a system that helps people better understand what personal data they are sharing, with whom, and to what extent; allows them to visualize the repercussions of this sharing; and allows them to make better choices and gain more freedom on when and how they share their data. This Comps project will begin to explore some of these areas.
The project
In this project, you will design and implement a privacy auditing system. The goals of the system are as follows:
- Broadly, help end users at a variety of technical proficiency levels better understand privacy issues, particularly as they relate to the sharing of their own data.
- Perform privacy audits of a user's personal data.
- Present the results of this audit in ways that end users at a variety of technical proficiency levels can understand and act upon, including the tradeoffs for tightening privacy in a given area.
- (if time permits) Allow mechanisms for the user to easily change his/her privacy settings in various domains.
The project entails the following activities:
- Research what metrics are currently used to define and evaluate privacy.
- Learn how privacy data is currently stored and structured: in what format is it stored? What are its contents? What metadata accompanies it? What standards currently exist for defining the contents and metadata?
- Figure out ways to trace the "data sharing chain". For instance, how widely is a particular piece of data shared? By how many parties, mediating agents, etc?
- Define your own set of privacy metrics by which to evaluate how much and how extensively a user's data is shared. For instance, one simple and possibly naive metric might define the "reach" of a piece of data as the number of distinct places in which a particular piece of data appears. Note: You don't necessarily have to reinvent the wheel here---it's perfectly acceptable to use/refine/modify metrics that you find in the literature.
- Audit all the ways and all the places (online, via apps, etc) where user data is shared, using the metrics you defined earlier (and, most likely, existing tools).
- Visualize and present the audit results in a way that can be understood by the end user.
- (if time permits) Determine mechanisms for modifying privacy settings on behalf of the user, without having the user go in and manually modify settings themselves. For instance, the system might present the user with an option to stop sharing and/or storing credit card data; if the user selects this option, the system would automatically find all places where credit card data is shared or stored and modify the settings to prevent this from happening. This might also entail using/integrating existing tools.
- (if time permits) Develop an appropriate interface through which users can operate the system.
- Conduct usability and feasibility tests on the system.
A major part of this project will be researching and discovering what's already been done in this space, and then building upon this work. This might mean integrating and adding to various existing tools, integrating some existing protocols, implementing an existing measurement framework, etc.
Deliverables
There are two main deliverables for this project:
- A paper and/or poster describing the data privacy metrics you defined and an assessment of how well these metrics work in practice.
- A command-line tool that performs a privacy audit of the personal data a person is sharing online, using the metrics you've defined. You will use this tool in your assessment of the metrics (see the first deliverable).
If time permits, you can do any of the following:
- Write up the paper/poster for submission to an appropriate academic conference.
- Expand upon the tool: develop an appropriate interface, expand upon the types of data the system audits, present visualizations of the data.
- Explore the idea of modifying privacy settings on behalf of the user.
Recommended experience
While there are no specific prerequisites for this project, some experience with ethics, computer security, and/or computer networks would be useful.
References/inspiration
- Serge Abiteboul, Benjamin Andre, and Daniel Kaplan. 2015. Managing your digital life. Commun. ACM 58, 5 (April 2015), 32-35. DOI=10.1145/2670528 http://doi.acm.org/10.1145/2670528
- Soren Preibusch. 2015. Privacy behaviors after Snowden. Commun. ACM 58, 5 (April 2015), 48-55. DOI=10.1145/2663341 http://doi.acm.org/10.1145/2663341
- Yuval Elovici, Bracha Shapira, and Adlai Maschiach. 2002. A new privacy model for hiding group interests while accessing the Web. In Proceedings of the 2002 ACM workshop on Privacy in the Electronic Society (WPES '02). ACM, New York, NY, USA, 63-70. DOI=10.1145/644527.644534
- Jennifer King, Airi Lampinen, and Alex Smolen. 2011. Privacy: is there an app for that?. In Proceedings of the Seventh Symposium on Usable Privacy and Security (SOUPS '11). ACM, New York, NY, USA, , Article 12 , 20 pages. DOI=10.1145/2078827.2078843