Formal work on hypothesis-driven models of recognition bring us face-to-face with the problem of ontological interoperability: the ontology used for generating models employed in hypothesis-matching must align with the schema of the databases being searched in order to control the computational cost of the model-building process. We describe preliminary work toward a system for hypothesis generation and corroboration. The system uses a plurality of ontological representation approaches plus application-specific biases to transform descriptions of probable candidate scenarios into descriptions of observable and deductively provable candidate scenarios, based on the available data sources. The transformation occurs at system setup time. The application biases come from the domain of threat detection.