DocMark: A Rule-based and Probabilistic Document Marking System
Small Business Information
P.O. Box 422, Trumansburg, NY, -
AbstractABSTRACT: The secure, automated dissemination of information across security domains requires accurate, complete, and standardized releasability markings. Existing sensors, information sources, and data processing tools do not all produce the required markings, nor can they always follow the latest version of standards such as the DoD Discovery Metadata Specification (DDMS). Further, contextual information and the current mission profile often determine the level of protection required for given data -- context that a sensor cannot possible have. The ATC-NY team will develop DocMark, a hybrid rule-based and probabilistic document marking system informed by the current mission profile. DocMark will fill in incomplete and non-standard releasability metadata for data objects in a repository, or for data provided via a web service API. In doing so, it will also provide additional provenance data, indicating the rules or algorithms used, confidence values, etc., that downstream processors can use in making release decisions. BENEFIT: DocMark"s application of releasability metadata rules and its ability to learn from existing marked data will dramatically speed the secure exchange of data -- especially when that data includes legacy information, sensor data, or other incompletely-marked data items -- enabling real-time or near real-time collaboration across security domains. By adding provenance information to any releasability metadata it supplies, DocMark will increase confidence in future automated or semi-automated release processes. The need for accurate, rule-based releasability markings extends beyond the DoD and Intelligence Communities into first responder groups (e.g., in releasing information to partners or to the public), health care (for medical records) and commercial enterprises that need to protect trade secrets yet share data with partners.
* information listed above is at the time of submission.