An important task for Homeland Security is the prediction of threat vulnerabilities, such as through the detection of relationships between seemingly disjoint entities. A common data structure used for this task is a semantic graph (a.k.a. an attributed relational graph). Such graphs encode various relationships as typed links between a pair of typed nodes. The node and link types are related through an ontology graph. Furthermore, each node has a set of attributes associated with it (e.g., "age" may be an attribute of a node of type "person"). In this talk, I will discuss the topological nature of real-world semantic graphs and describe two approaches for relationship detection in such graphs. The first approach uses probabilistic heuristics with quality bounds to efficiently find paths between two nodes. The second approach utilizes information-theoretic measures to discover and rank dependencies among relationships.
Presentation (PDF File)
Video of Talk (RealPlayer File)
Copyright. All Rights Reserved.