SBIR-STTR Award

While recent efforts in cross-domain information sharing have developed platforms and infrastructure for information sharing, they do not address the question of the content of the information that is being shared. This project will develop the latter capability: identifying information that is (i) relevant and (ii) sharable. This is a key problem in "just enough" data sharing, especially in crisis situations when there is not enough time available for human users to make this determination. We propose an innovative approach to semi-manual and automatic "information review" that can: (i) identify documents that should/shouldn't be shared, given the topical relevance of those parts to the users and task context, and (ii) identify parts of documents that should/should not be shared, based on security level of the domain and potential users. Our approach will enable development of an "intelligent information sharing assistant" (IISA) which will help cooperating groups share information in an operational setting without cognitive overload.

Keywords:
INFORMATION SHARING, INFORMATION ASSURANCE, ARTIFICIAL INTELLIGENCE, CASE-BASED REASONING, NATURAL LANGUAGE PROCESSING, DOCUMENT CATEGORIZATION, MACHINE LEARNING

The ability to identify and share information in a "just enough" manner is critical in defense, intelligence, and industry settings. Human analysts in both government and private sectors are overwhelmed by the sheer volume of data, compounded by the increasing complexity of information sharing policies. Unlike companies that provide infrastructure and platforms for information sharing, such as guards and servers, Enkia will provide software to analyze the content of what is being shared across such platforms. Using advanced artificial intelligence technologies, Enkia will be the first to offer adaptive products that can learn to identify sensitive and critical information without extensive hand-coding of information sharing policies. Our key insight is to leverage the same sheer volume of data that is overwhelming human analysts and use it to feed a data-hungry AI engine capable of learning from that data. Our technologies include natural language processing to analyze semantic content, case-based reasoning for document classification, and relevance feedback for automatic self-learning. This innovative combination of technologies will be deployed through a state-of-the-art service-oriented architecture, enabling usage via desktop plug-ins, server-based filters, and web service applications.

Keywords:
Information Sharing, Information Assurance, Content Review, Artificial Intelligence, Case-Based Reasoning, Natural Language Processing, Machine Learni