Private search in the real world

Encrypted search --- performing queries on protected data --- has been explored in the past; however, its inherent inefficiency has raised questions of practicality. Here, we focus on improving the performance and extending its functionality enough to make it practical. We do this by optimizing the system, and by stepping back from the goal of achieving maximal privacy guarantees in an encrypted search scenario and consider efficiency and functionality as priorities. We design and analyze the privacy implications of two practical extensions applicable to any keyword-based private search system. We evaluate their efficiency by building them on top of a private search system, called SADS. Additionally, we improve SADS' performance, privacy guaranties and functionality. The extended SADS system offers improved efficiency parameters that meet practical usability requirements in a relaxed adversarial model. We present the experimental results and evaluate the performance of the system. We also demonstrate analytically that our scheme can meet the basic needs of a major hospital complex's admissions records. Overall, we achieve performance comparable to a simply configured MySQL database system.