| Research in the field of reasoning with uncertainty has resulted in the powerful framework of probabilistic networks. Probabilistic networks nowadays are being successfully applied in a range of domains, such as for medical diagnosis. While considerable experience has been gained with building networks for the detection of disease in individuals, modelling the knowledge involved in detecting infectious diseases in a population has so far received little attention. Initial experiences have shown that the currently available framework does not suffice for this purpose, since the detection of such diseases involves reasoning not just about the signs observed in a selection of individuals but also about the disease patterns showing in the population. The focus of the projected research now is to enhance the framework of probabilistic networks to provide for reasoning with multiple levels of knowledge. More specifically, the fundamental research objectives are - to design an enhanced probabilistic-network formalism for modelling multiple levels of knowledge; - to design algorithms for effective reasoning with networks that capture multiple knowledge levels. The fundamental results will provide for the development of decision-support systems for complex problems that require reasoning about multiple levels of knowledge. More specifically, the results will allow the design of networks for the detection of infectious diseases in humans or in farm animals that are kept in groups. To study the practicability of the results in a real-life setting, a probabilistic network will be developed for the early detection of classical swine fever in pig herds. |
