Information about http://cups.cs.cmu.edu/soups/2006/posters/tarasewich-poster_abstract.pdf

Protecting the Privacy of Displayed Information …

Tags: based solutions, blinders, boston ma 02115, ccs, computer information science, conlan, current technologies, eyeglasses, gong, hci laboratory, huntington avenue, information privacy, laptop screens, mobile displays, northeastern university, owner information, privacy requirements, public spaces, web browser privacy,
Pages: 2
Language: english
Created: Fri Jun 2 18:49:44 2006
Display cached document
Page 1
image
Page 2
image
              Protecting the Privacy of Displayed Information
                                     Peter Tarasewich, Jun Gong, Richard Conlan
                  HCI Laboratory, College of Computer & Information Science, Northeastern University
                                360 Huntington Avenue, 202 WVH, Boston, MA 02115

                                           {tarase, gjoliver, kaige} @ccs.neu.edu

ABSTRACT                                                                    information while restricting strangers' access to this same
Current technologies allow users to access information in virtually         information. Privacy requirements will also vary based on the type
any public setting. This creates situations where sensitive                 of information, and on the preferences of the information's owner
information, both organizational and personal in nature, can be             (e.g., individual or organization) [e.g., 2].
seen and captured by nearby people and technology. Therefore,               Several hardware-based solutions have been explored to solve the
methods are necessary to ensure the privacy and security of                 problem of maintaining information privacy on mobile displays.
information displayed in public spaces. The authors have                    Privacy covers have been developed for laptop screens that
developed Web browser privacy blinders, which hide sensitive                provide a clear view of the screen's contents to the user but
information from view while leaving other information                       obscure the view to anyone looking at the screen from an angle.
unobscured. Research to date on this topic is reviewed, along with          Contents of screens can also be blurred, readable only through
current and future work                                                     devices such as special eyeglasses. While potentially valuable in
                                                                            protecting the privacy of information, these techniques may have
1. INTRODUCTION                                                             potential drawbacks in terms of 1) additional cost; 2) additional
Maintaining privacy in the mobile environment remains difficult             weight, bulk, and power consumption; 3) increased complexity;
because the context of a device or application can change rapidly           and 4) distortion or degradation of the displayed information,
and without notice. This is in sharp contrast to a fixed                    which could affect performance.
environment, like an office, where people can consistently control          Our ongoing research concerns the development and testing of
the way that information is handled to minimize the chance of               privacy blinders and related techniques. Privacy blinders [4]
divulging sensitive information to unauthorized parties. In an              mimic the use of yellow sticky-notes to cover parts of a larger
office, computer screens can be pointed towards a user such that            document so that they are not viewable by others. Blinders can be
other people cannot easily read them [1]. People are not                    used to provide a mixed display in which sensitive information is
intentionally careless when it comes to protecting information in           hidden (covered) but information not considered private is
public places, but normal human behavior makes it easy for                  displayed normally. If the user decides to view the sensitive
unsafe conditions to exist. For example, laptop computers are               information, they can temporarily remove the blinder. For
often used whenever and wherever needed or desired (e.g., in a              example, blinders on a tablet PC or PDA might be removed by
airplane). In this situation, the user can become more focused on           touching them with a stylus. When the stylus is removed from the
the task at hand rather than the fact that information might be             screen, the blinders reappear. It is also possible to create blinders
overseen or recorded by someone close by. While current                     that can only be removed with a certain gesture, thereby creating a
technology makes it easy to access information anywhere and                 level of security along with information privacy.
anytime, it does not concurrently provide adequate protection of
that information. Effective solutions to privacy protection                 Furthermore, privacy blinders can automatically respond to a
problems must not only be technically sound, but usable and                 predefined organizational and/or personal "privacy policy," which
understandable from a social perspective. Mobile users need                 specifies what types of information are covered under different
interaction methods that work well with multiple and varied tasks,          circumstances. An organizational policy might be dictated by the
and in environments that can change rapidly and potentially be              company a person works for, while a personal policy is
hostile. If this is not accomplished, users must accept tradeoffs           customized to a user's own comfort level and privacy
between the pervasive availability of information and the potential         requirements. This method can also account for user context
loss of privacy and security [1].                                           changes; if a person moves to a less public space, they might turn
                                                                            off the blinder feature and view all information without
While privacy is often maintained through methods (e.g.,                    obstruction. Context might also be taken into account
encryption) that keep data from being read by unauthorized                  automatically by the system. For example, a change in location
parties, this research looks at the relatively unexplored but equally       from a private office to a public meeting room might modify
important problem of maintaining the privacy of displayed                   privacy settings by design. This flexibility allows adaptation to the
information. Our overall goal is to create technically sound but            changing environment of the mobile device user.
practical methods of maintaining privacy of sensitive information
that is displayed in public and mobile environments. Any solution
must be resilient enough to work in any context (i.e., location and         3. BLINDERS TO HIDE INFORMATION
task independent), and ultimately adapt to changing contexts.               To date we have completed two pilot studies with privacy
                                                                            blinders. The first study [4] tested the basic concept of the
                                                                            blinders in terms of usability and effectiveness. We used a
2. BACKGROUND                                                               limited-function Mozilla Web browser prototype in a controlled
Privacy is valued and expected by most people to varying degrees.           laboratory study with each subject searching three "canned"
Usually an individual expects reasonable access to personal                 banking Web sites for specific information. Privacy blinders were


                                                                        1
displayed based on special HTML tags next to information that             removing a blinder, it might be set to drop its opacity level,
was defined to be sensitive in nature. The blinders were set to a         allowing the user to view the information through a translucent
predefined size, and appeared centered directly over sensitive            panel, but still discouraging onlookers who are at a distance from
information. The user could reveal information protected by a             the screen. When this is done, the software can also be compared
blinder in one of two predetermined ways. For one, the blinder            directly against hardware-based techniques for screen privacy.
disappeared when the stylus was moved over it. The blinder                One implementation difficulty has been determining what content
reappeared when the stylus moved away. In the second, the                 to cover. Currently the plug-in relies upon a simple matching
privacy blinder disappeared for a total of 10 seconds when a              paradigm to determine what to cover.
special stylus gesture was made, then reappeared.




                                                                             Figure 2. Configuration Dialog Box for Privacy Blinders.
                                                                          Since our pilot studies, we have also modified the plug-in to cover
                                                                          graphics as well as text, and continue to add to its functionality
                                                                          and ability for user customization. Our long-term goal is to
 Figure 1. A random webpage with privacy blinders enabled.
                                                                          establish an interface that allows the easy definition of a personal
In the second study [4], a FireFox Web browser extension with             "privacy policy" that can be used along with organizational
user-configurable privacy settings allowed users to browse                privacy settings as required (see [3] for a good discussion of
personal information from any Web site. Unlike the software from          developing privacy rules).
the first study, this extension did not simply look for predefined
                                                                          We are planning to run longitudinal field tests where participants
HTML tags. It preprocessed any Web page, located user-defined
                                                                          will be given a version of the software to run on their personal
sensitive information, placed user-customized blinders on top of
                                                                          devices. The software will automatically track information about
the specified content, and presented the resulting "blinded" Web
                                                                          the privacy settings and how often blinders appear on various
page to the user. Figure 1 shows a random Web page viewed with
                                                                          Web sites. The software will prompt the user for feedback after
privacy blinders. Properties of the privacy blinders that could be
                                                                          performing tasks with the browser. Versions of the privacy blinder
user-configured included 1) whether to group blinders, 2) to use
                                                                          software will also be created to run on PDA's and mobile phones.
fixed or variable blinder sizes, and 3) setting the blinder opacity
(transparency). Besides these blinder properties, the "privacy            Context data (such as location, co-location, and scheduled events)
policy" could also be customized through an interface shown in            might also be used to automatically ensure that a user is
Figure 2. Four classes of potentially sensitive content were              interacting with a mobile system in the safest possible manner,
supported. These were 1) monetary amounts (a number starting              and might increase privacy management effectiveness by shifting
with "$" or containing two decimal points); 2) email addresses            the burden of environmental awareness from user to system.
(abc@xyz.dom); 3) telephone numbers (e.g., xxx-xxx-xxxx); and
4) a user-specified delineated list of words and phrases that they        5. REFERENCES
wanted covered. All participants in this study responded                  1. Dourish, P., Grinter, R.E., Delgato de la Flor, J., and Joseph,
positively to using the privacy blinders, and felt that                      M. Security in the Wild: User Strategies for Managing
privacy blinders would be useful on a PDA or mobile                          Security as an Everyday, Practical Problem. Personal and
phone. Additional results of both studies can be found in [4].               Ubiq. Computing, 8, (2004), 391-401.
                                                                          2. Hawkey, K. and Inkpen, K. M. Keeping Up Appearances:
4. CONTINUING WORK                                                           Understanding the Dimensions of Incidental Information
While the original intent of the privacy blinders had been to                Privacy. Proc. of CHI 2006, (2006), 821-830.
protect sensitive personal and financial information from                 3. Karat, C.-M., Karat, J., Brodie, C., and Feng, J. Evaluating
onlookers, there are many alternative uses that we had not                   Interfaces for Privacy Policy Rule Authoring. Proc. of CHI
originally envisioned, including parental filters, workplace                 2006, (2006), 83-92.
privacy, highlighting, and color-coding of sensitive terms.
                                                                          4. Tarasewich, P., Gong, J., and Conlan, R. Protecting Private
Work continues on examining different sizes and shapes of                    Data in Public, Adjunct Proc. of CHI 2006, (2006), 1409-
blinders, alternate ways of placing and removing blinders, and               1414.
degrees of user customization One idea is rather than completely


                                                                      2