Implementing Δps (PS-Merge) Belief Merging Operator for Belief Revision

Oscar Chávez-Bosquez, Pilar Pozos-Parra, Jianbing Ma


Belief merging aims at combining information from multiple sources while belief revision studies strategies for retracting information in order to maintain consistency when the addition of new evidence to a belief base makes it inconsistent. An ordering of the sentences in the belief base is used to determine priorities among sentences so that those with lower priority can be identified and retracted. This ordering can be difficult to generate and maintain. To address this difficulty, in this paper we show how to generate automatically an ordering of the belief base sentences through the implementation of a belief merging operator. We extend the Δps (PS-Merge) belief merging operator in order to consider constraints, then we use this extension, called Δps-μ (Δps under constraints), as a strategy for belief revision. We treat new evidence as a constraint and apply the extended merging operator to obtain the revised belief base. We propose several properties of this operator when compared to other two belief revision operators solving four examples described as real-life scenarios. Finally we show a software prototype based on this approach, called Belief Reviser, freely accessible online.


Belief revision; belief merging under constraints, knowledge modelling; decision support systems; Δps (PS-Merge)

Full Text: PDF