АВТОМАТИЗАЦИЯ ДОКУМЕНТООБОРОТА С ПРИМЕНЕНИЕМ МАШИННОГО ОБУЧЕНИЯ И АНАЛИЗА ТЕКСТОВ

Abstract

The paper presents an overview of modern approaches to document workflow automation using machine learning and natural language processing techniques. Classical statistical text analysis models, distributed word representation methods, and transformer-based architectures are considered. A comparative analysis of these approaches is carried out in terms of classification accuracy, computational cost, and applicability in electronic document management systems. The study shows that context-aware models provide higher document processing quality, although they require increased computational resources. The main limitations of existing solutions are identified and promising directions for the development of intelligent document management systems are outlined.