A Two-Step Choice Privacy-Preservation Method for Check-in Data Publishing

Check-in data play an increasingly important role in mining users' behavioral patterns and personalized recommendation etc. However, check-in data abuse also rises users' concern about their privacy. Obfuscating one's check-in data to another's is an effective method. Existing approaches focus more on privacy preservation than on data utility. To the end, this paper proposes a two-step choice method. The method first clusters users into several clusters based on users' check-in data. Then the method chooses a cluster, named substituting cluster, which is the nearest to the cluster which the publishing user belongs to. The cluster which the publishing user belongs to is named original cluster. The substituting cluster and the original cluster should satisfy the predefined privacy threshold. The second step is to find a user in substituting cluster, named substituting user, whose check-in data are the most similar to the publishing user. Finally, the method replaces the publishing user check-in data by the substituting user's. Empirical evaluation shows that the proposed method can not only preserve the user's privacy effectively, but also preserve higher data utility.