Ontology Identification Problem In Computational Agents

Abstract

The Ontology Identification Problem is the problem of connecting different ontologies to the system’s goals in such a way that a change in the system’s ontology does not result in a change in its goal’s effect. My thesis is that the Ontology Identification Problem, which has so far been addressed as a single universal problem, can be seen as an umbrella term for a wide range of different problems, each of which has a different level of difficulty, and each requires different methods of approach, in order to overcome. One wide category of this problem is connected to granularity, where the changes in the model are connected to changes in the level of detail. Granularity issues can be divided into cases of simpler reductions, multiple realizability and incommensurability. Another wide area of the problem is related to context. Contextual problems can be divided into problems of environmental context and social context. Special cases of warrantless goals and perverse instantiation also have a direct bearing on the ability to solve ontology identification problems effectively.