PRIVACY PRESERVING MULTIPARTY COLLABORATIVE DATA MINING FOR MULTIPLE SERVICE PROVIDERS

We present a new multiple service provider model of operation for the Internet delivery of data mining services. So the collaboration becomes especially important because of the mutual benefit it brings. For this kind of collaboration, data's privacy becomes extremely important: all the parties of the collaboration promise to provide their private data to the collaboration, but neither of them wants each other or any third party to learn much about their private data. One of the major problems that accompany with the huge collection or repository of data is confidentiality. The need for privacy is sometimes due to law or can be motivated by business interests. Performance of privacy preserving collaborative data using secure multiparty computation is evaluated with attack resistance rate measured in terms of time, number of session and participants and memory for privacy preservation. Many anonymization techniques, such as bucketization and generalization, have been designed for privacy preserving publishing. Present work has shown that generalization loses considerable amount of information, especially for high-dimensional data. Bucketization, not clearly prevent membership disclosure and do not clear separation between sensitive and quasi-identifying attributes.