Angelia

Master's Thesis

December 2014

The goal of this project is to design and evaluate crowd- based techniques to highlight unusual and unexpected parts of privacy policies on web sites, so users do not need to spend a long time finding parts of privacy policies that might be of concern to them. The proposed method is to rank parts of privacy policies by comparing them and make people choose the one that they think to be more important to know. With this method, we obtained several lists of privacy policy statements sorted according to their importance. By combining this method along with other sorting techniques, we managed to find the method that is most effective, time and money-wise. We also found that we cannot consistently reach an absolute ordering of statement according to its importance with this method, but we deem that absolute ranking is not necessary since our main goal is to summarize the privacy policy. This finding suggests that crowdsourcing combined with ranking methods can be used to summarize long documents such as privacy policies.

crowdsourcing, privacy, privacy policies