Enhancing CMDB accuracy using AI-driven discovery and relationship mapping: A review

As digital infrastructures become increasingly complex and hybridized, the need for accurate, automated, and dynamic Configuration Management Databases (CMDBs) has never been more urgent. This review explores the application of Artificial Intelligence (AI) in enhancing CMDB accuracy through advanced techniques such as machine learning (ML), deep learning (DL), natural language processing (NLP), and graph neural networks (GNNs). Key AI capabilities include automated asset discovery, relationship inference, anomaly detection, and configuration drift management. By evaluating over a decade of academic and industry research, this paper provides a comprehensive taxonomy of AI models, a proposed architecture for implementation, and empirical comparisons of model effectiveness. The review also identifies prevailing challenges such as lack of data standardization, integration with legacy systems, and model explainability and proposes future research directions aimed at creating intelligent, self-healing, and transparent configuration management systems.