Advanced Approaches to KPI Management: Semantic Modeling and AI-Driven Forecasting

Key Performance Indicators (KPIs) play a critical role in monitoring and improving performance across domains such as business operations, manufacturing, healthcare, and finance. However, traditional approaches to KPI management often face significant challenges, including limited predictive accuracy, fragmented data sources, and poor interoperability. These limitations are particularly problematic in dynamic environments that demand real-time, data-informed decision-making. This thesis proposes an integrated framework for enhanced KPI management, combining semantic technologies with AI-driven forecasting methods. Part I of the research focuses on the development of ontology-based systems to extract, validate, and prioritize KPIs, improving data consistency and adaptability across organizational contexts. Part II introduces advanced forecasting models, leveraging Machine Learning (ML), Deep Learning (DL), and Transformer-based architectures to improve prediction accuracy and scalability. The proposed framework is implemented and evaluated using domain-specific datasets from retail, consumer business, and industrial manufacturing sectors, including the Rossmann sales dataset, Superstore dataset, and EPA greenhouse gas emissions data. These case studies demonstrate the framework’s ability to address forecasting, sustainability, and semantic modeling challenges in real-world applications. The results confirm the effectiveness of the framework in addressing key gaps in data representation, predictive modeling, and cross-domain adaptability. Overall, this research offers practical tools and theoretical contributions to build future-ready KPI systems that serve both organizational performance and broader societal goals. performance across domains such as business operations, manufacturing, healthcare, and finance. However, traditional approaches to KPI management often face significant challenges, including limited predictive accuracy, fragmented data sources, and poor interoperability. These limitations are particularly problematic in dynamic environments that demand real-time, data-informed decision-making. This thesis proposes an integrated framework for enhanced KPI management, combining semantic technologies with AI-driven forecasting methods. Part I of the research focuses on the development of ontology-based systems to extract, validate, and prioritize KPIs, improving data consistency and adaptability across organizational contexts. Part II introduces advanced forecasting models, leveraging Machine Learning (ML), Deep Learning (DL), and Transformer-based architectures to improve prediction accuracy and scalability. The proposed framework is implemented and evaluated using domain-specific datasets from retail, consumer business, and industrial manufacturing sectors, including the Rossmann sales dataset, Superstore dataset, and EPA greenhouse gas emissions data. These case studies demonstrate the framework’s ability to address forecasting, sustainability, and semantic modeling challenges in real-world applications. The results confirm the effectiveness of the framework in addressing key gaps in data representation, predictive modeling, and cross-domain adaptability. Overall, this research offers practical tools and theoretical contributions to build future-ready KPI systems that serve both organizational performance and broader societal goals.