TWC: Medium: Collaborative: Seal: Secure Engine for AnaLytics - From Secure Similarity Search to Secure Data Analytics

许多组织和个人依赖云来存储他们的数据和处理他们的分析查询。但这些数据可能包含敏感信息。用户不仅希望在云上隐藏其数据，还可能希望隐藏对其数据的分析查询、此类查询的结果以及来自云服务提供商的数据访问模式（这可能会受到内部或第三方的威胁）。本研究设计并实现了SEAL，一个用于云上大数据分析的安全引擎。SEAL使用安全的加密方案加密数据，但通过构建元应答数据库的新方法支持分析操作。SEAL是在一个允许指定不同级别所需安全性的安全框架内开发的。SEAL的设计特别探索了安全性和效率之间的权衡，为各种分析操作提供了具有不同可验证安全性和效率特性的解决方案。用户可以继续享受云计算所提供的好处，但现在不必担心丢失敏感信息，并且可以控制安全性和效率的权衡。

Many organizations and individuals rely on the cloud to store their data and process their analytical queries. But such data may contain sensitive information. Not only do users want to conceal their data on a cloud, they may also want to hide analytical queries over their data, results of such queries, and data access patterns from a cloud service provider (that may be compromised either from within or by a third party).This research designs and implements SEAL, a Secure Engine for AnaLytics over large data on a cloud. SEAL encrypts data using secure encryption schemes, but supports analytical operations through a new approach of building a meta-answer database. SEAL is developed within a security framework that allows specifying different levels of desired security. The design of SEAL in particular explores the tradeoff between security and efficiency, providing solutions with different provable security and efficiency features for a wide variety of analytical operations. Users are able to continue to enjoy the benefits a cloud has to offer, but now without the worry of losing sensitive information and with control over the security and efficiency tradeoff.