Reconstruction of possibilistic behavior systems

Abstract

Reconstructability analysis is viewed as a process of investigating the possibilities of reconstructing desirable properties of overall systems from the knowledge of the corresponding properties of their various subsystems. The reconstructability analysis consists of procedures for generating meaningful reconstruction hypotheses, procedures for the evaluation of the reconstruction hypotheses, and procedures for making various decisions regarding the acceptance of evaluated reconstruction hypotheses, generation of additional reconstruction hypotheses, termination of the analysis and the like. The paper discusses the evaluation of reconstruction hypotheses when the systems under consideration are possibilistic behavior systems. It is shown that a principle of maximum ambiguity, similar to the principle of maximum entropy for probabilistic systems, can be used for possibilistic systems. It is also shown that the unbiased (maximum ambiguity) reconstruction can be determined by a simple join procedure, in a similar fashion as for probabilistic systems. The join procedure for possibilistic systems turns out to be computationally simpler than the one for probabilistic systems. The paper also describes a general procedure for determining the reconstruction family.