Content area
The dissertation research is devoted to addressing a relevant scientific and practical problem – improving the efficiency of risk management in large-scale agile software development projects. This is achieved through the development of models and methods within an information technology framework designed to support goal-oriented risk management decision-making, assess the safety of artificial general intelligence during the requirements engineering phase, and enhance the effectiveness of risk identification and managerial decision-making.
The relevance of the research topic on risk management has been substantiated in the context of the current development of information technologies, where scalable agile approaches have become a core methodology in software product development. The dissertation establishes the connection of the research with broader scientific themes, defines the object, subject, and research methods, and highlights the scientific novelty and practical significance of the obtained results. The study also presents the validation of the proposed solutions and their dissemination through scientific publications.
Chapter 1 analyzes the current state of theory and practice in software project management. It includes a review of development life cycles in scalable agile methodologies and identifies key factors influencing risk manifestation. The research demonstrates the limitations of existing models in addressing the specific nature of scalable agile development and justifies the need for a more adaptive model that combines methodological flexibility with advanced analytical tools. It was established that such factors as team interaction complexity, unpredictability of requirements, and technical constraints significantly affect project outcomes.
Chapter 2 is devoted to the study of the systems approach to risk management. A comprehensive analysis of risk taxonomy has been conducted, encompassing the identification of internal and external risks. Mechanisms for risk prevention and mitigation of their potential consequences have been proposed. Logic-algorithmic models for risk management have been examined. Furthermore, an analysis of the multiplicative effect of risk factors resulting from their interaction under conditions of scalable software development has been carried out.
Chapter 3 introduces and describes the goal-oriented software risk management model, which includes project goal setting, risk identification, obstacle modeling, risk assessment, and mitigation planning. The approach is unique due to its systematic use of project goals as a foundation for risk management and integration of risk-handling activities into requirements engineering. The model aligns risk management processes with project success objectives derived from development components. Defining goals contributes to risk reorganization and simplifies risk management. Recognizing risks enhances goal clarity. The “component-element-factor” hierarchy enables holistic identification and classification of goals and risks, facilitating the construction of the goal-risk model.
Chapter 4 expands the capabilities of the model by incorporating AGI safety aspects into software development. It evaluates potential AGI threats such as data breaches, algorithmic malfunctions, and other risks. A method for AGI risk assessment using AHP and optimization via GA is proposed. The approach’s effectiveness is demonstrated through optimal risk management strategy generation and integration into development processes.
It has been established that the integration of the Analytic Hierarchy Process (AHP) and Genetic Algorithms (GA) within the framework of a goal-oriented risk management model for software development has enabled the formation of a comprehensive approach to addressing critical risk management tasks. In particular, it facilitated the implementation of justified risk prioritization based on objective expert evaluations, the optimization of management strategies considering their effectiveness and resource constraints, as well as the development of an integrated fitness function that incorporates both qualitative and quantitative risk characteristics. This approach enhances the substantiation of decision-making and aligns the risk management process with the achievement of strategic project objectives in the context of large-scale agile software development.
The developed mathematical foundation for combining AHP and GA enables the integration of qualitative and quantitative risk evaluation methods. The proposed risk assessment matrices support a structured approach to prioritization based on multiple weighted criteria. GA-based optimization algorithms demonstrate effectiveness in solving complex multifactor risk management tasks where traditional methods fall short. The integration of AHP and GA improves calculation accuracy and accounts for dependencies among various risks. Examples of calculations and practical application show the method’s high efficiency across fields such as project management, financial analysis, logistics, and production processes. The integration of AHP and GA significantly reduces risk levels and supports optimal management decisions. The proposed approach is universal and can be adapted to other management tasks, offering broad prospects for practical implementation. Future research may focus on algorithm improvements, adaptation to dynamic system changes, and integration into real business processes.
The proposed unified fitness function incorporates risk prioritization and resource constraints, enabling efficient allocation of resources to minimize critical risks. The combination of risk weights and available constraints allows the method to adapt to real project environments and optimize management decisions. The function's performance is confirmed through examples that demonstrate high accuracy and speed in solving multicriteria problems. This method can be applied across various industries for rational risk and resource management, with further improvement and expansion potential.
Scientific novelty of the results obtained.
For the first time:
– A software risk management model has been proposed that integrates goal levels, obstacles, risk assessment, and mitigation/elimination actions into the requirements engineering process.
– Methods have been developed to account for risks associated with the implementation of artificial general intelligence at the early stages of the lifecycle in large-scale agile software development.
Improved:
– The systematization and classification of the main types of risks inherent in large-scale agile software development projects, taking into account the possible consequences of their realization, which – unlike existing approaches – enables goal-oriented risk management and the implementation of appropriate actions during the requirements engineering phase.
– The risk management approach by developing a fitness function that integrates risk weights and resource constraints, enabling the optimization of decision-making in a multicriteria environment and ensuring flexible adaptation to real project conditions.
Further developed:
– The Analytic Hierarchy Process (AHP), through its adaptation for multicriteria risk analysis in large-scale agile projects, which increases the accuracy of risk assessment and enhances the justification of management decisions.
The practical significance of the obtained results lies in enhancing the effectiveness of risk management through the use of the proposed risk management model, which enables the identification, assessment, and mitigation of risks at the early stages of software development. This ensures the resilience of projects even in the dynamic environment of scalable agile development.
The implementation of the proposed solutions leads to a reduction in project costs and risks, as the use of multi-criteria analysis mechanisms (such as the Analytic Hierarchy Process) facilitates optimal decision-making in risk management. This, in turn, contributes to lower problem-solving expenses and minimizes the likelihood of project failures.
The use of the goal-driven model allows adaptation to current development challenges through the integration of artificial general intelligence (AGI) risk assessment into the requirements engineering process, which helps mitigate threats associated with the adoption of new technologies – an especially relevant issue for modern high-tech projects.
The proposed models and methods support scalability, as they are suitable for risk management in both small projects and large-scale scalable initiatives, making them versatile tools for use in IT companies.
The scientific and practical results of the research have been implemented in the activities of software development companies, in particular, Information Technologies of Trade LLC, Apptimized Operations, UA Technics/PerSys Medical (Ukraine), and in the educational process within the «Information Technology Design» bachelor’s program, specialty 122 «Computer Science» of Sumy State University, as evidenced by the relevant acts and implementation certificates.
The results of the dissertation research were presented and discussed at national and international conferences: 6th International Conference on Design, Simulation, Manufacturing: The Innovation Exchange (DSMIE-2023), 2023, High Tatras, Slovak Republic; 63rd International Scientific Conference on Information Technology and Management Science of Riga Technical University (ITMS), 2022, Riga, Latvia; 64th International Scientific Conference on Information Technology and Management Science of Riga Technical University (ITMS), 2023, Riga, Latvia; Conference "Integrated Computer Technologies in Mechanical Engineering – 2022. Synergetic Engineering", 2022, Kharkiv, Ukraine; XXI International Scientific and Practical Conference "Project Management in the Development of Society", 2024, Kyiv, Ukraine.