Modeling a probabilistic ontology for Maritime Domain Awareness

Situational awareness and prediction are essential elements of information fusion. Both involve various types of uncertainty and require a sound automated inferential process. Probabilistic ontologies support uncertainty management in se-mantically aware systems, and facilitate modular, interoperable systems. This paper describes the process of developing a probabilistic ontology for a Maritime Domain Awareness application. The ontology was created to support identification of ships behaving suspiciously enough to be declared ships of interest. The original model was expanded in two ways: to provide reasons for declaring a ship as being of interest, and to include individual crew member associations. The latter is achieved by supporting inferences about a person's close relations, group associations, communications, and background influences to assess his likelihood of having terrorist links.