Abstract: Information discussing and trades across a website boundary are desirable for a number of application needs. We suggested o-MPPI to deal with the requirements of differentiated privacy protection of multi-term phrases inside a PPI system. To better of our understanding, o-MPPI is first focus on the issue. O-MPPI guarantees the quantitative privacy protection by carefully controlling false positives inside a PPI and therefore effectively restricting an attacker’s confidence. To deal with the difficulties of efficient secure o-MPPI construction, our core idea would be to draw a line between your secure part and non-secure part within the computation model. We minimize the secure computation part whenever possible by exploring various techniques. Hence separated the complex NLP computation in the MPC part so that the costly MPC within our o-MPPI construction protocol only is applicable to a simple computational task, thus optimizing overall system performance. In the style of o-MPPI, we recognized a collection of challenging problems and suggested novel solutions. For just one, we formulated the quantitative privacy computation being an optimisation problem that strikes an account balance between privacy upkeep and check efficiency. We addressed the cruel problem of secure o-MPPI construction within the multi-domain information network which lacks mutual trusts between domain names.

Keywords: Information sharing, Privacy protection, o-MPPI construction, Secure computation, Multi-domain information, Mutual trust, Privacy preservation.