Similarity Measures for Persona/Human Networks

Under the ONR Large Tactical Sensors Networks (LTSN) program, we have demonstrated a semantic wiki application, SWIM, capable of dynamically assembling and organizing data from disparate sources into a semantic graph conformant to an ontology. Using SWIM, incoming information is linked, where possible, to existing entities in the semantic graph, with linkages subject to user validation via the semantic wiki. For this SBIR Phase I we propose to investigate and demonstrate expanding SWIM capabilities in two areas: a) iterative feature set matching algorithms that will hypothesize probabilistic persona-persona matches as a mechanism to suggest merges of existing entities. This capability will improve identification of individuals based on partial or fragmentary information. b) subgraph matching algorithms that assess similarity of evolving social network or persona role or behavior identification against canonical group structure templates. This capability will improve the association of persona with groups. We will also assess methods to structure concepts representing motivators of behavior, leading to future research connecting individuals, groups and ideals to specific predicted behavior patterns.