METHODS OF APPLYING ARTIFICIAL INTELLIGENCE FOR OPTIMIZING BIG DATA PROCESSING IN MODERN INFORMATION SYSTEMS

Abstract

The purpose of the study is to develop applied approaches to intelligent management of computational resources and optimization of data processing in environments with highly dynamic workloads. The focus is placed on the interaction between machine learning algorithms and mechanisms of resource allocation and task scheduling. The research aims to identify relationships between the structure of computational flows, system parameters, and query processing speed. The methodology is based on a combination of economic and mathematical modeling, time series analysis, and experimental testing of algorithms in distributed systems. Classification methods, adaptive learning techniques, and fuzzy logic are applied to construct management models. A comparative analysis of the efficiency of different load balancing strategies is conducted, along with an assessment of query execution time in streaming environments. The obtained results demonstrate that the integration of hybrid algorithms reduces processing delays and increases system stability under variable workloads. A relationship between the accuracy of resource prediction and the level of computational node utilization is identified. A decrease in query execution time is observed when adaptive task scheduling models are applied. It is proven that the use of intelligent management mechanisms reduces computational costs without compromising system performance. The practical significance lies in the possibility of applying the proposed approaches in high-tech enterprises operating with large volumes of data. The results can be used to improve information systems, optimize logistics processes, and enhance the efficiency of digital platforms. The proposed models are suitable for implementation in real-time systems with limited resources. The scientific novelty consists in the formation of an integrated approach to resource management that combines machine learning methods, forecasting, and data processing optimization. The results expand existing approaches to the design of intelligent management systems and provide a foundation for further applied research in the field of digital technologies.