CSR: Small: Collaborative Research: Multi-party Collaborative Data Access

点击翻译按钮获取中文摘要

As the richness and volume of data maintained by a wide variety of businesses and organizations increases, there is a growing need to derive useful information, not only from data owned by individual parties, but also across parties. The current data sharing practice is dominated by carefully negotiated, mostly secret, one-on-one agreements between organizations, and is seriously limited. It does not scale as the number of databases and organizations increase, cannot be formally analyzed for unintended information leaks or other violations, and may require unnecessary data exposure. This project explores a collaborative multi-party data sharing mechanism that allows safe and convenient sharing of databases in order to enable large-scale collaborative data analytics and knowledge extraction applications. Multi-party data sharing involves many challenges that the project attempts to address. First, it is necessary to have an intuitive specification of access control policies for data access across parties and an automated mechanism to translate those into a concise set of access rules while resolving conflicts, specification inadequacies, and other problems. Second, we need efficient and scalable implementation of allowed queries in spite of complex access restrictions. Third, it is necessary to handle both the traditional relational databases, its variants, and the emerging graph databases. Finally, we need mechanisms to efficiently check the integrity and completeness of the data supplied as answers. By addressing these challenges, the research hopes to establish the foundations of multiparty data sharing. Many of the formalizations in this research will exploit Satisfiability Modulo Theories (SMT) based methods for testing assertions and for resolving conflicts. With tremendous amount of data being generated by both cyber and cyberphysical systems, this project will help devise practical mechanisms to share data in a wide variety of scenarios and thereby accelerate the pace of data driven innovation applied to information systems serving the society.