Smart and private social activity invitation framework based on historical data from smart devices

Modern social networks bring people together and help facilitate the organization of various group activities. The rapid development of smart wearable devices has also made feasible the extrapolation of their owners' activity habits. Inspired by the recent work by Ai et al. [2], we design a smart and private social activity invitation framework based on historical data from smart devices. Our paradigm aims at helping users organize group activities in a smart and efficient way while finding compromises to satisfy all involved parties. Compared with Ai et al.'s work [2], our framework is more realistic, whereby users report their personal information to the app server, which is used to provide organizing services to registered members. The app server, however, is untrustworthy and could be motivated by factors such as advertising revenue. Therefore, the app may advertise itself by providing aggregate statistical information about current users to attract new users. This creates a dilemma between the existing users' concerns about personal privacy and the app developers' agenda. Our framework ameliorates this conflict by securing existing users' information under a state-of-the-art privacy concept - differential privacy - guaranteeing quality services to existing users, while also allowing the server to give informative answers to new potential users. In addition, the proposed framework encourages less active or isolated users via a new method based on perturbed graphs. Our simulation results demonstrate that the proposed framework performs well.