Employing Data Mining to Create an Improved Information Intelligence System

We propose an innovative combination of text analysis and data mining techniques in the design of an information intelligence-based program management system. By drawing on our extensive experience, we have devised a novel approach to translating free andfield-delineated text into quantitative data suitable for use with a wide range of data mining techniques. In particular, our system (FlexiMiner), will support the application of text preprocessing, term discovery, concept formation, and automated textsegmentation techniques in the encoding of text documents. The resulting quantitative data can take on a variety of forms depending on the goals of the analyst and the available computational resources. FlexiMiner will offer an improved capability toexplore material needs documents and open source R&D abstracts, as well as discover logical linkages between these two sets of information. Finally, the ability to do cross-field pattern discovery and free-text analysis will offer more flexibility thanthat offered by the Technology Opportunities Analysis System. Phase I research and development of a proof-of-concept limited prototype will build on our existing data mining toolkit, IKODA, and lay the groundwork for the Phase II implementation of a fullyfunctional common tool suite for both text and quantitative data mining.The US military is not alone in its need for advanced text mining tools. The proposed system would be very valuable to venture capitalists, medical researchers, and engineering designfirms.