A STUDY OF APPROACHES TO DATA PROCESSING AND FACTORIZATION IN GEOINFORMATION SYSTEMS

Abstract

Purpose. To conduct a comparative analysis of existing approaches to geospatial data processing in geographic information systems and to determine the potential for applying algebraic factorization to the analysis of multidimensional environmental data.

Methodology. A systematization and comparative analysis of approaches to geospatial data processing was performed, specifically including methods of distributed computing, geometric modeling, matrix factorization, multidimensional statistical analysis, spatio-temporal graph neural networks, and approaches to integrating data from various sources. The evaluation was based on the following criteria: data type, presence of a spatial component, consideration of temporal dependencies, application of algebraic factorization, integration with GIS, scalability, interpretability of results, computational complexity, and suitability for environmental tasks. Four groups of approaches were considered: infrastructural and theoretical methods, algebraic factorization methods, statistical methods, and intelligent algorithms, which allowed for an assessment of their effectiveness in processing spatial, spatiotemporal, and multidimensional geospatial data.

Findings. It has been established that neural network models effectively account for complex spatiotemporal dependencies but are characterized by high computational complexity and low interpretability. It is shown that existing approaches do not provide a universal combination of methods of algebraic factorization, spatiotemporal modeling, and the specifics of environmental geospatial data.

Originality. Modern approaches to geospatial data processing have been systematized from the perspective of applying algebraic factorization, and unresolved aspects of its integration into environmental geoinformation systems have been identified. The feasibility of developing factorization methods for multidimensional data that account for spatiotemporal dependencies has been substantiated.

Practical value. The results can be used to create new methods of algebraic factorization in environmental geoinformation systems, in particular for environmental monitoring, decision support, and the analysis of multidimensional geospatial data.