Martin Berzell

The thesis investigates how the notion of ‘ontology’ has been used in the field of medical informatics and knowledge representation. Partly to investigate what an ‘ontology’ can be said to represent and what requirements we can have on a good ‘ontology’. The author studies the already existing medical terminologies and ‘ontologies’ to elucidate what theories they are based on. The terminological theories of Eugen Wüster and his legacy in medical informatics are studied. It is noted that terminological theories handling linguistic entities are not suited for describing and representing medical theories, since these are assumed to refer to the real world, which consists of more than linguistics entities. In order to find a metaphysical theory in accordance with the world view that medical theories describe, the author turn to the critical realism of Karl Popper, Roy Bhaskar and Ilkka Niiniluoto. These theories, taken together with the metaphysical theories regarding universals of David M Armstrong and Ingvar Johansson, are used as a basis to find out what an ‘ontology’ can be said to represent, and what criteria and requirements we can have on a good ‘ontology’. Among the requirements presented in the thesis are stability, interoperability and the requirement that a good ‘ontology’ must be in accordance with our best available theories. Finally, it is discussed how these requirements and criteria can come into conflict with one another, and how one should reason when handling these trade-offs. The author emphasises the importance of including the medical expertise in the process of creating ‘ontologies’, in order to produce as useful and relevant ‘ontologies’ as possible.