Dealing With Poorly Mapped Entities in Adaptive Object-Oriented World Modeling

As a prerequisite to supporting situation assessment and awareness tasks, information about considered situations must be adequately represented in respective support systems. A meaningful representation can be provided by world modeling systems like the Object-Oriented World Model (OOWM). The OOWM models the current state of an application domain based on observed information while relying on domain knowledge for adding semantics to the model. This domain knowledge is usually designed a priori by human experts. Yet, in complex real-life domains, the occurrence of a priori not considered domain entities is likely during system operations. Therefore, the OOWM has to be able to deal with a potentially open world -by adapting the domain model in reaction to current and past circumstances. To this end, an approach for adaptive knowledge management has previously been proposed. Extending the approach, this contribution introduces a holistic process for managing and dealing with unforeseen entities supporting model adaptation and details a method for acquiring new concept definitions based on poorly represented entities.