Knowledge acquisition is known as a major "bottleneck" in the development of expert systems. Problems increase, if multiple experts have to be considered, because they usually have different views of the domain, e.g. a manufacturing system, i.e. the usually use different terminologies to describe the domain and do not agree in their decisions. In the proposed approach, terminological problems are overcome by relating the terminologies of the experts to the terminology of a basic model using machine learning techniques. Based on this relationship, a translation among the terminologies of the experts using a two-step procedure becomes possible. To acquire the knowledge of the experts, using machine learning, especially using Inductive Logic Programming is discussed. Finally, approaches for discovering and overcoming of conflicts in the behaviour of the experts are presented.
展开▼