Elevating Semantic Search by Marrying Social Network and Content Analysis Techniques Navy SBIR FY2010.2 Sol No.: Navy SBIR FY2010.2 Topic No.: N102-180 Topic Title: Elevating Semantic Search by Marrying Social Network and Content Analysis Techniques Proposal No.: N102-180-0222 Firm: Stottler Henke Associates, Inc. 951 Mariner's Island Blvd., STE 360 San Mateo, California 94404 Contact: Terrance Goan Phone: (206) 545-1478 Web Site: www.stottlerhenke.com Abstract: The military's ability to exploit streams of heterogeneous information sources has not kept pace with the enterprise's ability to generate them. In order to make substantial progress in enhancing the warfighters' situational awareness, new approaches are required to automatically link related information. We propose to develop this capability by tapping the synergy between existing text analysis and Social Network Analysis (SNA) techniques. Our overall approach can be viewed as "pearl-culturing", in which our system will first utilize an ensemble of existing text analytics tools to identify clusters of strongly related documents. These tools will include (among others) our own proven technology for identifying documents that were either derived from a common information source or provide details regarding the same event. Our system will then employ ontology-based extraction tools to extract entities and relations from these tight clusters (treating them as if they were a single document). This approach will allow us to utilize existing technology to create more complete entity profiles than has been previously possible. Finally, SNA metrics will be used to quantify and qualify the relationships among documents, entities, and concepts. Ultimately this technology will support improved semantic search as well as SNA across the enterprise. Benefits: The proposed technology will have a wide range of applications in areas where analysts face serious challenges in collecting and fusing data hidden in a sea of electronic information. While support entity profiling and SNA within the military and the Intelligence Community offer obvious commercialization opportunities, we also see the potential to improve enterprise search applications more broadly. Specifically, we believe that a technological leap could be achieved that would match the significance of Google's PageRank scheme. PageRank works so well on the Web because of its heavily hyperlinked structure - something missing within most enterprises. The proposed technology can provide an analogous (and potentially more powerful) linking structure to the enterprise: providing semantic search capabilities that are supported by enterprise ontologies but not dependent on them. Return