User Insisted Redistribution of Belief in Hierarchical Classification Spaces

摘要

Where users and agents, each with their own world model and expertise, work together it is essential to interpret both their beliefs correctly. It is therefore important to keep track of the differences of opinion that occur in such a way that it is understandable for both the agents as well as the user. This paper proposes a generic and flexible way or the user to interact with agents using a integrated world model. To enforce the user’s opinion a User Preference Redistribution rule (UPR) is proposed. Through a realistic numerical example we show the validity of this model and the new UPR in contrast to other belief conditioning rules.