Methods and algorithms of the spelling control and correction on the basis of fuzzy qualifier with MIMO-structure

The paper shows results of developing the conceptual principles and methods in construction of hyper semantic net for natural languages spelling control and correction on a basis of neural networks and methods of fuzzy logic at the expense of expert knowledge and account of uncertainty. The ways of formalization are offered for linguistic variable and parameters, represented quantitatively and qualitative, and for performance of fuzzy sets membership functions. The operability of net including various nodes of graph model is tested. The realization of net is carried out on the basis of model with MIMO-structure, combining in itself property of the neuro-fuzzy qualifier and fuzzy production system.