Kalman-like filtering and updating in a possibilistic setting

1 IRIT Université Paul Sabatier – 31062 Toulouse Cedex 4 – France. Email: {benferhat, dubois, prade}@irit.fr Abstract: This paper proposes a qualitative counterpart of Kalman filtering in the possibilistic logic setting. It corresponds to a type of updating involving a prediction step followed by a revision step. This is compared with updating operations based on imaging in the sense of Lewis. Imaging is reconsidered in the perspective of a generalized view of Kalman filtering where it appears as a particular case of Kalman filtering. A syntactic counterpart of qualitative filtering is given in terms of weighted knowledge bases.