Online Query Answering with Differential Privacy: a Greedy Approach using Bayesian Inference

Data privacy issues frequently and increasingly arise for data sharing and data analysis tasks. In this paper, we study the problem of online query answering under the rigorous differential privacy model. The existing interactive mechanisms for differential privacy can only support a limited number of queries before the accumulated cost of privacy reaches a certain bound. This limitation has greatly hindered their applicability, especially in the scenario where multiple users legitimately need to pose a large number of queries. We propose a greedy algorithm using Bayesian statistical inference for online query answering, which minimizes the use of privacy budget when answering each query with a utility requirement. The key idea is to keep track of the query history and use Bayesian inference to answer a new query using previous query answers if the inference result satisfies the utility requirement; Otherwise the query is answered with the minimal privacy budget corresponding to its utility requirement. The Bayesian inference algorithm provides both optimal point estimation and optimal interval estimation given the observations of the query. We show that our approach maintains lower privacy budget usage, answers more queries, achieves a longer system life span and provides more accurate estimations than traditional approach through extensive experiments on different real-life data sets.