S.S. Krylov1, A.O. Zhukov2, A.D. Shabliy3, V.A. Sudakov4

1, 3-4 Moscow Aviation Institute (National Research University) (Moscow, Russia)

2 Expert and Analytical Center (Moscow, Russia)

1 compgra@yandex.ru, 2 aozhukov@mail.ru, 3 alexey.shabliy@gmail.com, 4 sudakov@ws-dss.com

This article addresses the pressing issue of software decomposition in the context of plug -in systems. The key objective is to minimize the amount of functionality delivered as part of a sof tware product that is useless to a specific customer. The relevance of this work stems from the need to address issues typical of plug -in architectures, such as licensing conflicts and functional anomalies arising from the presence of unused code in the system. The primary goal of the study is to substantiate the feasibility of reducing the amount of useless functionality by optimizing the distribution of the source code files implementing it among individual plug-ins. To this end, a formal graph-analytical model of the problem domain has been developed. The model integrates key entities: functional requirements, source code files, plug-ins, and the relationships between them–the traceability of requirements to files and the dependencies between the files themselves. Based on this model, an optimization problem has been formulated whose objective function is aimed at minimizing the total number of implement ed but useless requirements across all considered delivery packages. The solution to the problem is to find an optimal file distribution matrix for plugins. A practical test of the proposed method was conducted using a real -world open -source project, "meta -configurator." During the codebase analysis, 161 functional requirements and 124 source code files were identified, traceability and dependency matrices were constructed, and 10 configuration options were defined. Two algorithms were used to solve the optimization problem: reinforcement learning and a genetic algorithm. Computational experiments showed that increasing the number of plugins with optimal file distribution does indeed significantly reduce the amount of useless code in the distribution. A comparative analysis demonstrated the absolute superiority of the genetic algorithm over the RL method in both computational speed and the quality and stability of the resulting solutions. Thus, the results of the study confirm the proposed hypothesis. The main contribution of this study lies in the successful ad aptation of classical decomposition principles (fr om microservice architecture) to plug -in systems and the proposal of a new objective metric focused on the customer value of functionality rather than technical aspects. Prospects for further research lie in en hancing the complexity of the model by incorpo rating cost characteristics of requirements and the use of specialized solvers for integer programming problems.

Krylov S.S., Zhukov A.O., Shabliy A.D., Sudakov V.A. Software decomposition in plug-in systems. Dynamics of complex systems. 2026. V. 20. № 2. P. 24−37. DOI: 10.18127/j19997493-202602-03 (in Russian).

1. Abgaz Y., McCarren A., Elger P., Solan D., Lapuz N., Bivol M., Jackson G., Yilmaz M., Buckley J., Clarke P. Decomposition of Monolith Applications Into Microservices Architectures: A Systematic Review. IEEE transactions on software engineering. 2023. V. 49. № 8. P. 4213–4242. DOI: 10.1109/TSE.2023.3287297.
2. Rudrabhatla C. Impacts of Decomposition Techniques on Performance and Latency of Microservices. International Journal of Advanced Computer Science and Applications. 2020. V. 11. № 8. P. 19–24. DOI: 10.14569/IJACSA.2020.0110803.
3. Velepucha V., Flores P. A Survey on Microservices Architecture: Principles, Patterns and Migration Challenges. IEEE Access. 2023. V. 11. P. 88339–88358. DOI: 10.1109/ACCESS.2023.3305687.
4. Al-Debagy O., Martinek P. A Microservice Dec omposition Method Through Using Distributed Representation of Source Code. Scalable Computing: Practice and Experience. 2021. V. 22. № 1. P. 39–52. DOI: 10.12694/scpe.v22i1.1836.
5. Hafiz Hasan M., Hafeez Osman M., Indriaty Admodisastro N., Sufri Muhammad M. From Monolith to Microservice: Measuring Architecture Maintainability. International Journal of Advanced Computer Science and Applications. 2023. V. 14. № 5. P. 857 –866. DOI: 10.14569/IJACSA.2023.0140591.
6. Blinowski G., Ojdowska A., Przybylek A. Monolithic vs. Microservice Architecture: A Performance and Scalability Evaluation. IEEE Access. 2022. V. 10. P. 20357–20374. DOI: 10.1109/ACCESS.2022.3152803.
7. Faustino D., Gonçalves N., Portela M., Rito Silva A. Stepwise migration of amonolith to amicroservice arc hitecture: Performance and migration effort evaluation. Performance Evaluation. 2024. V. 164. DOI: 10.1016/j.peva.2024.102411.
8. Lin J., Sayagh M., Hassan A.E. The Co-evolution of theWordPress Platform and Its Plugins. ACM Transactions on Software Engineering and Methodology. 2023. V. 32. №
9. Article 19. P. 1–24. DOI: 10.1145/3533700.
10. Lima I., Cândido J., d'Amorim M. Practical detection of CMS pluginconflicts in largeplugin sets. Information and Software Technology. 2020. V. 1 1
11. Article 106212. P. 1–13. DOI: 10.1016/j.infsof.2019.106 2 1
12. Software decomposition in plug-in systems Dynamics of complex systems / Dinamika slozhnykh sistem, V. 20, № 2, 2026, p. 24−37 37 1
13. Viet Nguyen H., Kästner C., Nguyen T.N. Exploring variability-awareexecution for testingplugin-based web applications. ICSE 2014: Proceedings of the 36th International Conference on Software Engineering. 2014. P. 907–918. DOI: 10.1145/2568225.2568300. 1
14. Wintersgill N., Stalnaker T., Heymann L.A., Chaparro O., Poshyanyk D. The Law Doesn’tWork Like a Computer: Exploring Software Licensing Issues Faced by Legal Practitioners. Proceedings of the ACM on Software Engineering. 202 4. V.
15. Article 40. P. 882 –905. DOI: 10.1145/3643766. 1
16. Xu W., He H., Gao K., Zhou M. Understanding and Remediating Open-Source License Incompatibilities in the PyPI Ecosystem. 2023 38th IEEE/ACM International Conference on Automated Software Engineering. 2023. P. 178–190. DOI: 10.1109/ASE56229.2023.00175. 1
17. Neubauer F., Bredl P., Xu M., Patel K. Design, Implementation, and Evaluation of a Meta Configuration Tool Using a Schema-to-UI Approach. Datenbank-Spektrum. 2024. P. 1–9. DOI: 10.1007/s13222-024-00472-7. 1
18. Neubauer F., Pleiss J., Uekermann B. Data Model Creation with MetaConfigurator. Datenbanksysteme für Business, Technologie und Web (BTW 2025). Gesellschaft für Informatik. Student Track. 2025. P. 933–944. DOI: 10.18420/BTW2025-60. 1
19. Neubauer F. Data modelcreation with MetaConfigurator. Institute for Visualization and Interactive Systems. University of Stuttgart Universitätsstraße 38 D–70569 Stuttgart. P. 1–83. DOI: 10.18419/opus-15126. 1
20. Neubauer F., Uekermann B., Pleiss J. Data Model Creation with MetaConfigurator. Zenodo. 2025. DOI: 10.5281/zenodo.14981537. 1
21. Ladosz P., Weng L., Kim M., Oh H. Exploration in deep reinforcement learning: A survey. Information Fusion. 2022. V. 85. P. 1 –22. DOI: 10.1016/j.inffus.2022.03.003. 1
22. Abel D., Barreto A., Van Roy B., Precup D., Van Hasselt H. A definition of continual reinforcement learning. 2023. DOI: 10.48550/arXiv.2307.11046. 1
23. Song Y., Wei L., Yang Q., Wu J., Xing L., Chen Y. RL-GA: A Reinforcement Learning-based Genetic Algorithm for Electromagnetic Detection Satellite Sc heduling Problem. Swarm and Evolutionary Computation. 2023. V. 7
24. Article 101236, P. 1 –29. DOI: 10.1016/j.swevo.2023.101236. 2
25. Huang S., Kanervisto A., Raffin A., Wang W., Ontañón S., Fernand Julien Dossa R. A 2 CisaspecialcaseofPPO. 2 0 2 2. D O I: 10.48550/arXiv.2205.09123.