Accurately Matching Requirements to Solutions through the Accumulation and Exploitation of Background Knowledge

We propose an innovative yet pragmatic approach to the matching of information warfare requirements to solutions. The proposed system concept, Resolve, will be fashioned from an eclectic set of proven technologies with the goal of both improving search precision and recall, and dramatically reducing the effort required to find matching solutions. Resolve will exploit a unique capability to extract requirement-descriptive terminology from proposal documents that, through a series of processing steps, can fuel the autonomous collection of supplemental requirement-related documents from the Web and local repositories. This additional textual material will then be used to facilitate a query expansion process that will in essence mediate between a user's query and the body of indexed solutions (mitigating the terminology mismatch problem). Resolve will offer further improvements by utilizing a proven context-enhanced process that exploits user feedback and pseudo-feedback to autonomously scour a topical search space, reducing the need for countless query reformulations. By exploiting a number of existing technological components Stottler Henke will be able to make rapid progress, yielding a limited yet fieldable prototype in Phase I. This prototype and its evaluation will provide a solid foundation for the complete implementation of Resolve in Phase II and its eventual commercialization.