User Opinion based Trust Value Prediction for Online Social Network

Abstract

Online social networking has caused profound changes in the way people communicate and interact. Preserving information privacy is indispensable in such social applications as the shared information would be sensitive. The issue becomes more challenging because of participation of multiple parties on the same shared data. Here propose an efficient trust collection based data sharing technique to allow or disallow the shared resources considering the authorization requirements of all the multiple parties. A logical representation of the proposed data sharing technique is prepared to analyze the privacy of data before sharing to public users. A user in this system is associated with a few trusted users that were selected from the user’s friends. When the user wants to share information to the account, the service provider sends information to the user’s trustees. The user must obtain at least k (i.e., recovery threshold) threshold values from the trustees before being directed to public share. Considering that a user continually posts data items in an OSN, here model the threshold selecting problem as a sequential decision-making problem. More specifically, we formulate the problem as a multi-armed bandit problem and apply the upper confidence bound (UCB) policy to solve the problem. This shows that dynamically adjusting the threshold according to the UCB policy can lead to a higher payoff than using a fixed threshold.