CleverCOMSRL: Implementation of an AI Computer-Aided Design System in the Context of the Cognitive Science Paradigm for the Research Training Process

The CleverCOMSRL system aims to enhance the computer's ability to approximate the end user. It automates the process of algorithm construction and program formation using functional semantic networks. The problem is solved by automating cognitive activity with the help of artificial intelligence formalisms. The deep learning network integrates functional systems, initiating their self-learning process. The system incorporates an integral long-term memory. To retain data for multiple iterations within a short time frame, an LSTM network is used in conjunction with Transformers class generative networks. To automate knowledge, role-based process frames and aggregate frames are merged into a unified structure. Prototype frames are transformed into instance frames and combined into a network of frames, which represents the technical model, when selecting units and their connection scheme. The CleverCOMSRL dialogue systems consist of the system core and management tools. The functional content follows a modular principle and is a collection of programmed variable modules. The content management system is based on principles of artificial intelligence within the context of the cognitive science paradigm. The study uses a complete Turing test, machine vision for object perception, and robotics tools for object manipulation and spatial movement. The CleverCOMSRL intelligent system aims to improve the efficiency and quality of research training. The experimental study presents validated variance results for each of the analyzed groups with an error probability of p < 0.005. The method of dimensionality reduction proposed here retains 60–80% of the original correlation information, while considering the intellectual development level of students.