Finding Reliable Subgraphs from Large Probabilistic Graphs

Consider information search or discovery in a large graph of concepts and their weighted relationships. The user initiates a query by specifying some concepts ("search terms"), and wishes to obtain other concepts and relationships that connect the search concepts. An application example is in analysis of biological information, conveniently represented as a graph of biological concepts and their relations. A search engine we envision would allow a life scientist to query for connections between a gene and a phenotype, for instance, to find information supporting a hypothesis, or to help discover new hypothesis.