Outsourced Privacy-Preserving C4.5 Algorithm over Arbitrarily Partitioned Databases

Many companies want to share data for datamining tasks. However, the privacy and security concerns become a bottleneck in the data sharing filed. The secure multiparty computation (SMC)-based privacy-preserving data mining has emerged as a solution to this problem. However, traditional SMC solutions are inefficient. We introduce the method of outsourcing to reduce the computation cost of user's side. In order to preserve the privacy of the sharing data, we propose an outsourced privacy-preserving C4.5 algorithm on arbitrarily partitioned databases based on the outsourced accountable computing for finding frequent itemsets (OAC-FFI) and the outsourced weighted average computing (OWAC) protocols. As a result, we show that our method can achieve similar result with the original C4.5 decision tree algorithm, but also preserve the privacy of the data. We prove that there is a sublinear relationship between the computational cost of user side and the number of participating parties.