COMPARISON AND IMPLEMENTATION OF SECRET SHARING SUITABLE FOR LIFE LOG SYSTEM

Recently, the application of data mining to big data has been attracting much attention. However, the huge data to which data mining is applied include users' personal information. Thus, data mining that preserves the privacy of the user is now an active area of research; in particular, secure multiparty computation (MPC), which allows encrypted data to be calculated without decrypting it, is being studied. In this paper, first we present two methods broadly classified as "secret sharing" and "homomorphic encryption", which can perform secure multi-party computation. In the secret sharing method, the burden of the required storage capacity falls on users who try to manage their own data. On the other hand, in the homomorphic encryption method, a user or a cloud system that tries to operate secure multi-party computation bears the large calculation burden. The GCCE method, which we proposed at GCCE 2013, can, however, resolve these problems. Then, we evaluate each method in terms of the required storage capacity and amount of calculation. Finally, we show the results of implementing GCCE in a specific application.