Differentially private data publishing: Interactive setting

Abstract

Interactive settings operate on various aspects of the input data, including transactions, histograms, streams and graph datasets. This chapter discusses publishing scenarios involving these types of input data. In interactive settings, the privacy mechanism receives a user’s query and replies with a noisy answer to preserve privacy. Traditional Laplace mechanisms can only answer sublinear of n queries, which is insufficient in many scenarios. Different mechanisms are discussed to fix this essential weakness.