A prototype of online privacy-preserving questionnaire system

The author participates in an ongoing project at Software Technology and Applications Competence Center (STACC). The aim of this project is to develop and test an online questionnaire system that is able to preserve participants’ privacy. In other words, the goal is that the original answers to a survey never leave user’s computer. However, we should still be able to get the aggregated results of all users. To perform privacy-preserving computations on secret shared data, we use the Sharemind framework [1]. This framework requires three independent parties to perform as data miners to guarantee privacy of the data. Current state of web-based privacy-preserving data mining using Sharemind framework is described in more detail in [2] and [3]. We use the same overall architecture as introduced in these works. [2] describes two possible technologies to implement the given architecture — JavaScript and Adobe Flex framework. Since JavaScript is unable to make cross-domain connections due to security restrictions, one would have to use a proxy server and Public Key Cryptography to guarantee privacy. In this case, a malicious proxy server would still be able to fake the public keys and gain access to the original data. When using Adobe Flex framework, the questionnaire application would be able to make direct HTTPS connections to three independent miners and the system does not depend on the proxy server. Thus, Public Key Cryptography is still used but the trust relationships between the end user and miners are established directly.