Privacy-preserving friend recommendations in online social networks

Online social networks, such as Facebook and Google+, have been emerging as a new communication service for users to stay in touch and share information with family members and friends over the Internet. Since the users are generating huge amounts of data on social network sites, an interesting question is how to mine this enormous amount of data to retrieve useful information. Along this direction, social network analysis has emerged as an important tool for many business intelligence applications such as identifying potential customers and promoting items based on their interests. In particular, since users are often interested to make new friends, a friend recommendation application provides the medium for users to expand his/her social connections and share information of interest with more friends. Besides this, it also helps to enhance the development of the entire network structure. The existing friend recommendation methods utilize social network structure and/or user profile information. However, these methods can no longer be applicable if the privacy of users is taken into consideration. This work introduces a set of privacy-preserving friend recommendation protocols based on different existing similarity metrics in the literature. Briefly, depending on the underlying similarity metric used, the proposed protocols guarantee the privacy of a user\u27s personal information such as friend lists. These protocols are the first to make the friend recommendation process possible in privacy-enhanced social networking environments. Also, this work considers the case of outsourced social networks, where users\u27 profile data are encrypted and outsourced to third-party cloud providers who provide social networking services to the users. Under such an environment, this work proposes novel protocols for the cloud to do friend recommendations in a privacy-preserving manner --Abstract, page iii