An integrated approach to document regrading with classification assistants

Stottler Henke proposes an integrated approach to document classification and dissemination termed InfoAssist. We will build upon the state-of-the-art technologies developed in Phase I to provide intelligent dirty word detection methods as well as techniques based on historical documents that help detect secure content in documents slated for dissemination. These techniques allow document authors and reviewers to make use of the release and publication history of all content in a particular document when assigning security classifications. Users will benefit from integrated tracking of a documents classification while they combine content from multiple sources, all without modification to their existing workflow. Furthermore, InfoAssist will provide an opportunity to solicit the input of document reviewers regarding enhancements to existing policy. This input can be carefully considered and—if suitable—gradually adopted into a semi-automated dissemination process through a rigorous testing and evaluation procedure. The Phase I development of InfoAssist provides a solid foundation of intelligent classification algorithms upon which we will build the Phase II implementation of the InfoAssist system. This system will integrate seamlessly with existing reviewing and dissemination tools, such as the Work-flow Enforcement Server, reducing the cost of adoption and ensuring immediate rewards.