Undersea Domain Multi-level Security Data Miner Navy SBIR FY2016.1 Sol No.: Navy SBIR FY2016.1 Topic No.: N161-050 Topic Title: Undersea Domain Multi-level Security Data Miner Proposal No.: N161-050-0678 Firm: InferLink Corporation 2361 Rosecrans Avenue, Suite 348 El Segundo, California 90245 Contact: Steven Minton Phone: (310) 341-2446 Web Site: http://www.inferlink.com Abstract: In this project, we propose to develop an approach for intuitive multi-level security data mining and user support, which we call MultiMiner. We address four key requirements in this project. First, we provide a data management structure at all classification levels by leveraging the existing security infrastructure of WebTAS, an accredited, government-off-the-shelf framework providing secure data management. Second, we propose a data-mining method that can operate across security levels through an approach where data is compressed at lower security levels and then passed upwards, so that the mining process benefits from access to the greatest amount of available information at each level. Third, we include an intuitive user interface enabling searching and displaying information, based on GeoStratus, which provides users with the ability to view three-dimensional plots, workflow, shapes and layers. Our approach integrates GeoStratus' capabilities with the taxonomic and faceted navigation search enabled within WebTAS' Topic Builder, an operational semantic search capability that provides intuitive access to heterogeneous data sources. Lastly, to achieve auto completion of reports we propose an approach based on hierarchical "Active Templates" so that reports can be easily filled in with a minimum of manual effort. Benefits: Advances in computer technology are continuing to introduce new technologies at a breakneck pace. With these new technologies comes more and more data. Users who need to discover relevant content are often adrift in a sea of possibly-relevant data. This problem is exacerbated in military environments where there can be multiple systems, and multiple networks, so that operators are forced to spend considerable time hunting for the data they need. This project is intended to result in a system that will make it significantly easier for users to find and display information that they need, as well as help them generate reports based on data from multiple sources. The system will be capable of data mining in a multi-level security environment, with an intuitive user-interface. In addition to the clear need for this technology in military environments, we have identified commercial cyber security applications where this technology would be valuable. Return