Moskal'kova N. Plausible inference methods on the basis of representation of the attributive knowledge models in semantic networks

The dissertation is devoted to the development of methods and software of inductive inference and inference by analogy on the basis of representation of the attributive knowledge models in semantic networks. The modification of the network construction algorithm and the method of the inference by analogy in pyramidal networks are developed. The structural-attributive knowledge models as a development of the attributive knowledge models are offered. The library of knowledge processing operations and methods of inference by analogy are developed for them. The offered algorithms have been realized in the system of analysis and discovery of the regularities CONFOR and the program complex "Analogy", which are the powerful tools of the task solving in discrete environment.