Algebraic models for data and knowledge representation in modern database management systems

The article discusses algebraic data and knowledge representation models in modern database management systems. It is shown that despite the effectiveness of the relational model in storing large volumes of structured information, its capabilities are limited for expressing machine learning algorithms. In this regard, new approaches are proposed based on advanced algebraic models that allow formalizing the architecture and operations of neural networks in SQL. Methods of hybridization of SQL and GPU computations, application of specialized operators, combining data processing and analysis stages are considered. The results confirm the high efficiency of the developed solutions for intelligent analytics.