Development of a Neural Network System for Creating Optimal Personal Computer Configurations Based on User Tasks

This paper proposes an approach to automated personal computer (PC) configuration generation using neural network technologies. The key stages of system development are considered: collecting data on user tasks and hardware specifications, designing a neural network architecture, and training the model based on multi-factor optimization. A hybrid architecture combining convolutional neural networks for text query analysis and graph neural networks for component compatibility is proposed. A comparison with traditional methods (expert systems, recommendation algorithms) is provided. Experimental results demonstrated a 15–20% reduction in configuration costs while maintaining target performance.