Negated Min-Based Possibilistic Networks

Possibilistic networks are important tools for reasoning under uncertainty. They are compact representations of joint possibility distributions that encode available expert knowledge. The first part of the paper defines the concept of negated possibilistic network which will be used to encode the reverse of a joint possibility distribution. The second part of the paper proposes a propagation algorithm to compute a possibility degree of each event in the negated possibilistic network. Our algorithm is based on the use of a junction tree associated to the initial graphical structure.