Resumen
A modification of the hybrid intellectual learning environment based on the concept of a knowledge based system of the production type, (analytical component) and neural network technologies for learning scenario planning (a synthetic component) is proposed. Specifically, attention is primarily given to the problem of formalization of production-type knowledge at the conceptual level (expert level) in the form of a solution graph that allows us to describe the fuzzy logic of the learning scenario planning, taking into account the individual characteristics of the learner. At a sufficiently deep level, the relationship between the infological model and the conceptual representation of knowledge in the form of a set of product rules for the analytic component of the learning environment is described. The technique of triggering rules (activation of associated events) in the process of logical inference is designed with taking into account the need to provide information to the user, testing the level of its assimilation and determining the outcome of the event. Two-level representation of events (compact and detailed) is proposed to provide a certain flexibility to the system. In terms of modification of the synthetic component of the learning environment to a more detailed form, the scheme for transforming the graph of solutions into an equivalent (according to the logic of "reasoning") feed-forward neural network (multi-layer perceptron) is brought. Neuron activation functions are defined which provide tact-by-tact management of the learning process and fixing its current state in the working memory of the intellectual learning environment. It is planned to memorize the full sequence of passed states in order to ensure the possibility of returning to them for a second consideration of the material that has been passed. The inexpediency of controlling any parameters of the neural network due to the lack of such networks is justified (all the functionality of the network is determined by its structure).