NobleBlocks

Research indexJournal article

КЛАСИФІКАЦІЯ МІНЕРАЛІВ У БАГАТОВИМІРНОМУ ПРОСТОРІ ГЕОХІМІЧНИХ ДАНИХ

Гамій

SCIENTIFIC PAPERS OF DONNTU Series “The Mining and Geology”8 May 2026

NobleIDark:/48914/w1/20261105/A292C177DOI10.31474/2073-9575-2026-1(35)69-76

Paywalled

The full text is at the publisher

You can read the abstract here, or open it at the publisher.

Read at the publisher

Abstract

Purpose To develop a methodology for the classification of minerals based on their chemical composition and physicochemical characteristics using statistical analysis and compositional data analysis methods with the cosine similarity metric for the identification of gypsum and minerals with similar geochemical properties. Methods A multidimensional dataset was collected and processed, comprising chemical composition, physicochemical properties, structural characteristics, crystal structure, Mohs hardness, transparency, specific gravity, optical properties, refractive index, chemical elements, lanthanides, transition metals, heavy elements, actinides, superheavy elements, chemical groups, as well as other parameters such as molar mass, molar volume, and calculated density, covering 3,110 minerals, including gypsum (CaSO₄•2H₂O) and other analogous sulfates and carbonates. The study was based on an analytical approach combining statistical data analysis with the cosine similarity metric. Results The conducted study was aimed at developing a methodology for mineral identification based on their chemical composition and physicochemical characteristics. As a result, a comprehensive approach was developed, and on the basis of a constructed statistical model, mineral classification was implemented with gypsum (CaSO₄•2H₂O) identified as a reference class and the degree of similarity of other minerals to it determined. The obtained results demonstrated a high structural and chemical affinity of gypsum to anhydrite, thenardite, glauberite, bassanite, kieserite, and other sulfate minerals, confirming the correctness of the formed feature space and the effectiveness of the applied computational methods. The proposed statistical model provides a quantitative assessment of mineralogical similarity, enables the automatic formation of a hierarchy of related minerals, and minimizes the subjectivity of expert-based classification. The developed approach demonstrates universality and can be applied to problems of mineralogy, geochemical forecasting, deposit analysis, and the development of mineral raw material recognition systems. Scientific novelty. A comprehensive methodology for mineral identification is proposed, combining a cosine similarity metric with statistical analysis methods adapted to the analysis of multidimensional geochemical data of a compositional nature. The proposed statistical model enables a transition from traditional descriptive mineralogical classification to an objective similarity metric that forms a continuous space of mineral affinity. The use of gypsum as a reference mineral for constructing an analogue-search model is substantiated, which made it possible to identify a group of structurally and genetically related minerals without prior expert grouping. A statistical classification model has been developed that is capable of scaling with the growth of mineralogical databases and being augmented with new features, including spectral, crystal-chemical, and thermodynamic parameters. The proposed approach establishes a methodological foundation for the development of mineral identification systems and for tasks related to forecasting the composition of mineral raw materials. Practical significance. A statistical model has been developed that enables accelerated identification of minerals, particularly gypsum, as well as the detection of minerals with similar characteristics. The proposed approach is intended for use in optimizing geological exploration activities, processing laboratory analysis results, preliminary classification of mineral raw materials, and supporting decision-making in mining-geological and materials science practice. The method makes it possible to significantly reduce dependence on expert judgment and interpretation, while increasing the objectivity of determining mineral associations.

Discussion

More

Suggest a correction or report