RULE-BASED HYBRID INTELLIGENT LEARNING ENVIRONMENT IMPLEMENTATION

Abstract

Learning is considered as an intelligent process the development scenario of which, with an individual approach to the learner, is not known in advance. The scenario is built during learning material studying and it largely depends on the intellectual abilities of the the student, his preparation level, psychological traits, learning environment conditions and other factors. Learning process is treated as a movement in state-space representation, where state transition is an event linked with output to terminal logically finished piece of learning material with grading (if needed). In this regard, we introduced an implementation of the hybrid intellectual learning environment shell which is based on rules and neural network inference and may be tuned to different subject areas using knowledge base. The author describes the implementation principles of main subsystems, their lingware, information provision and software. Solution graph is introduced to represent expert (teacher) knowledge on the conceptual level. The graph allows us to compactly describe logic of reasoning which is used while learning scenario planning. In the knowledge acquiring subsystem transformation algorithm is envisaged from solution graph to rule base (for analytical part of the learning environment) or to equivalent straightforward neural network (for synthetic part). The author describes a mechanism of the scenario tact by tact formation governed by neural network with possibility of two-level representation of the learning material (compact and detailed) and returning to previous system states to refresh past materials. As a prospect for further development, the problem of developing a methodology for creating a source graph of a solution for subject areas was mentioned.

Keywords

• Knowledge based system
• production rules
• neural network
• hybrid intellectual learning environment
• individual learning scenario