Prediction Problem’s Solution for the Finite Possibilistic Model of Expert Knowledge Streams

The solution of the prediction problem is presented for the finite possibilistic modelling [3,4,6,7]. A recurrent variant of finite possibilistic models is considered. In this variant, we define the regularization condition for constructing a quasi-optimal estimator of fuzzy transition operator (FTO). We construct the discrete recurrent extremal fuzzy process with possibilistic uncertainty, the source of which is an expert knowledge stream on the states of some evolutionary complex system. Expert knowledge stream is created on the basis of Dempster-Shafer temporalized belief structure. Based on the non-probabilistic utility theory the example is presented to illustrate the prediction problem solution for expert knowledge streams. Key-Words: possibilistic modelling, fuzzy process, expert knowledge stream, Dempster-Shafer belief structure, prediction problem