METHODOLOGY FOR CREATING AN INTEGRATED RESEARCH ENVIRONMENT BASED ON JUPYTER NOTEBOOK USING NEURAL NETWORKS

doi:10.26565/2222-5617-2025-42-09

Yu. V. Lytvynov V. N. Karazin Kharkiv National University, 4 Svobody Sq., 61022 Kharkiv, Ukraine https://orcid.org/0000-0002-4639-6771
H. V. Klymenko V. N. Karazin Kharkiv National University, 4 Svobody Sq., 61022 Kharkiv, Ukraine https://orcid.org/0009-0000-3658-849X

DOI: https://doi.org/10.26565/2222-5617-2025-42-09

Keywords: Jupyter Notebook, neural networks, Python

Abstract

This paper presents a methodology for creating an integrated research environment for experimental data processing based on Jupyter Notebook with the use of artificial intelligence tools, particularly generative neural networks. Jupyter Notebook is a leading platform among researchers due to its flexibility, broad library ecosystem, and ease of integration with various analytical tools. Its primary purpose is to enable researchers to create customized software solutions by writing code in more than 40 programming languages.

Traditionally, the development of research tools in Jupyter Notebook requires coding in the Python programming language. However, this can pose challenges for specialists who lack deep programming skills. With the rapid development of generative neural networks, a unique opportunity has emerged to create small, specialized programs for personal use without significant immersion in coding. Moreover, this approach not only simplifies the creation of applied software but also significantly accelerates the acquisition of programming skills, lowering the entry barrier into the development profession.The paper reviews the key capabilities of Jupyter Notebook, provides a brief overview of its interface, and offers basic explanations of the principles of program creation using neural networks. A crucial step in building research tools is the formulation of software functionality and interface design. Given that code generation is performed using neural networks, particular attention is paid to prompt engineering principles for effective code generation and the creation of applications for automating the processing of data in various formats. Specific examples of functional module development are presented, demonstrating the adaptability of neural network models to address typical experimental data processing tasks. The article is intended for experimental researchers seeking to enhance their analytical capabilities using modern neural network technologies while avoiding complex programming.

Downloads

Download data is not yet available.

References

1. J. Perkel. Nature, 563, 145 (2018). https://doi.org/10.1038/d41586-018-07196-1
2. T. Rule, B. Birmingham, D. Clayton, et al. PLoS Comput Biol 15(7) (2019). https://doi.org/10.1371/journal.pcbi.1007007
3. S. White. arXiv, 2302.11382 (2023). https://doi.org/10.48550/arXiv.2302.11382
4. Laria Reynolds, Kyle McDonell. arXiv, 2102.07350 (2021). https://doi.org/10.48550/arXiv.2102.07350
5. M. Chen, J. Tworek, H. Jun, et al. arXiv, 2107.03374 (2021). https://doi.org/10.48550/arXiv.2107.03374