DAIS: Maintaining Dynamic Aggregated Information Security

Federated search tools are of great value to users with timely information needs, but little attention has been directed towards solving the security issues that arise with these tools in the context of multiple security domains. Search results and the requested documents must be cleared for the requesting user before they can be presented; however, it is extremely costly to involve human reviewers in this information retrieval process. Thus, new automated techniques are needed to accurately identify the classification level of search results. Such techniques must do more than identify the classification of individual phrases, since high-security knowledge may be exposed by combining multiple portions of low-security content that, when combined, reveal classified information. These information aggregates must be detected and tracked automatically before a dangerous amount of the aggregate is released. We present a solution that integrates text analysis techniques with search technologies to enable automatic dissemination of information at multiple security levels based on user and system access rights as well as security markings. We then extend this solution to incorporate aggregate tracking and we propose methods to learn about and reason with external sharing to further reduce the chances of security leaks through content aggregation. BENEFIT: Many areas of the commercial sector have a pressing need for tools designed to improve information dissemination efficiency and security. The recent development of the Health Insurance Portability and Accountability Act (HIPAA) of 1996 and Sarbanes-Oxley Act of 2002 have made these requirements more pressing than ever. Our proposed approach, DAIS, incorporates state-of-the-art text analysis and information management technologies with federated search tools to manage information retrieval, both of individual documents and aggregates, in an efficient and secure manner.