Assessing the innovation activity of an organization using artificial intelligence systems

This article presents a set of scientific and methodological principles for assessing an organization’s innovation activity, developed using a process-based approach with the integration of artificial intelligence (AI) technologies. The research’s novelty lies in the development of a comprehensive methodological framework combining a process-based approach, multi-level barrier analysis and AI technologies to assess innovation activity in specific institutional settings. For the first time, an integrated scenario-based management mechanism and the architecture of an intelligent analysis system, adapted to the challenges of sanctions pressure and logistical isolation, are proposed. The study aims to address the current problem of adapting innovation assessment tools to the conditions of regions with specific institutional barriers, using the Republic of Crimea as a case study. The object of the study is the innovation activity of organizations in the Republic of Crimea operating under sanction restrictions and logistical isolation. The subject of the study is a process-based approach to assessing an organization’s innovation activity using AI systems. The methodological basis of the work includes a process-based approach, institutional analysis, SWOT analysis, mathematical modeling and scenario planning. The empirical base includes data on regional organizations, regulatory acts and statistical data. The study addressed a set of interrelated tasks. Institutional and digital barriers to innovation activity were identified and systematized at the micro-, meso- and macrolevels, establishing their synergistic nature. An original multi-level model for diagnosing these barriers was developed, enabling the identification of obstacles at all stages of the innovation cycle. A matrix for assessing the impact of various factors was proposed and tested, revealing the dominant role of economic constraints at the microlevel. A scenario-based management mechanism was constructed, including three development options («Adaptive Growth», «Focus and Efficiency» and «Antifragility and Localization»), ensuring the adaptability of the strategy. An architecture for an intelligent analysis system based on machine learning methods (Random Forest, XGBoost, NLP) was developed and a detailed scheme for integrating AI into assessment activities was designed. The practical significance of the work was confirmed by the testing of elements of the developed methodological framework. The results are intended for public authorities to formulate regional innovation policies and for company management to improve the effectiveness of innovation management in the face of uncertainty and technological challenges.