|
|
| Overview | Coloring Privacy | |
With the development of visualization tools and techniques, visual query systems have become more and more appealing to data analysts. The visual support with data representation and graphical interface could enhance data learning and analyzing process. Take OpenPaths as an example, locations and paths of people could be viewed on maps to provide insights to researchers in their corresponding fields such as human behavior. Many applications like OpenPaths have given a reliable privacy setting with encryption or anonymization. We may wonder if it is possible to push the limits of data utilization further via more flexible private setting and better visual representations. Below are the problems that we are working on for our research.
- We would like to explore the relationship between the parameters used in privacy-preserving mechanisms and the precision loss in the process of visualization due to map resolution limit or color range. With the consideration of visual attributes, less amount work could be needed to achieve the same privacy expectation and to obtain better visual utilities such as coloring contrast.
- Visualization could leak sensitive information about individuals. Hence, there must be guidelines to ensure privacy. Rather than showing attacks on visualization, we will enforce privacy to allow data owner to have better knowledge on their privacy and hence encourage them to share their data and participate in research.
- Moreover, simple views or previews of data with stricter privacy setting, such as the aggregate map of all OpenPaths location data, could be shown to data analysts at the first stage, so that the data analysts could have a better idea on the data type and relevance before requesting further information from the users. We could explore different levels of privacy settings and possible previews of data to enhance better usage of the data.
- Last but not least, each individual’s data only get unencrypted and used with his or her approval to participate in the research to ensure data security. On the other hand, we may like to explore if we could query on the encrypted data directly without decryption to obtain some useful analysis.
- Prof. Ashwin Machanavajjhala(advisor)
- Xi He
- Coloring Privacy Resarch project for the course, 'COMPSCI590.03, Privacy in Mobile-Social System', advised by Ashwin Machanavajjhala. In this project, one of the important features in visualization has been explored with respect to ε-differential privacy. Details can be found here.
- More ...
 |