W.E. Wolfengagen¹, L.Yu. Ismailova², S.V. Kosikov³
¹ˑ²National Research Nuclear University MEPhI (Moscow, Russia)
³LLC "JurInforR" (Moscow, Russia)
¹jir.vew@gmail.com, ²LYIsmailova@mephi.ru, ³kosikov.s.v@gmail.com
The widespread use and increasing quality requirements for statistical analysis highlight the problem of its integration with conceptual domain modeling methods. Existing statistical analysis systems are typically poorly connected to classical data models, such as relational ones, and require users to have deep expertise in statistics and programming. This creates a high entry barrier and limits the use of these methods by domain specialists who are not statistics experts. Furthermore, analytical procedures are often performed in isolation from the conceptual data model, complicating the interpretation of results and the systematic planning of experiments.
The aim of this work is to develop an approach for integrating conceptual domain modeling tools and statistical analysis methods. This involves creating a unified conceptual model that describes both the domain objects and the statistical methods applied to them. This approach is designed to lower the entry barrier by generating recommendations for using statistical methods, providing context-dependent interpretation of results, and automating the analysis process.
The paper presents an integrated model based on a polymorphic type theory. A conceptual model of statistical analysis methods has been developed, which includes data typing and formal descriptions of key procedures. The model enables the generation of a statistical experiment plan and explanations for the obtained results within a single conceptual framework.
The proposed solution has high practical significance as it is oriented toward users who are not professional statisticians. It helps to overcome language and professional barriers associated with using complex statistical packages. The integrated approach facilitates the interpretation of results and promotes a more informed application of statistical methods. The solution is particularly valuable for the educational process, offering methodologically refined examples and the ability to work with individual datasets for better comprehension of the material.
Wolfengagen W.E., Ismailova L.Yu., Kosikov S.V. Integration of conceptual domain modeling and statistical analysis tools. Information-measuring and Control Systems. 2025. V. 23. № 5. P. 61−66. DOI: https://doi.org/10.18127/j20700814-202505-06 (in Russian)
- Wolfengagen V., Ismailova L., Kosikov S. When and Where Conceptual Maths Equals to Conceptual Modeling: Reasons for Using in Cognitive Modeling. Studies in Computational Intelligence. 2024 V. 1130 LNCS. Q4. P. 973−979.
- Wolfengagen V., Ismailova L., Kosikov S., Slieptsov I., Dohrn S., Marenkov A., Zaytsev V. Semantic configuration model with natural transformations. Cognitive Systems Research. 2024 V. 83. Q1.
- Wolfengagen V., Ismailova L., Kosikov S. Inferable Methods for Identifying Objects. Studies in Computational Intelligence. 2024 V. 477 SCI. Q4. P. 435−441.
- Wolfengagen V., Ismailova L., Kosikov S. Logical-Applicative Computing Based on Type Theory. Studies in Computational Intelligence. 2024 V. 477 SCI. Q4 P. 426−434.