Object-centric process mining enables the analysis of complex business processes involving multiple interacting objects, such as in inventory management or e-business workflows. However, long object lifecycles often result in convoluted event logs, obscuring actionable insights and challenging process mining techniques. This paper introduces a segmentation framework that decomposes object-centric event logs into focused, analytically tractable units called segments. We propose segmentation strategies based on object relationships, event types, and temporal intervals, and integrate them with machine learning techniques for anomaly detection and correlation analysis. Implemented in the OC-PM tool, our approach empowers users to uncover localized patterns and optimize process execution. A case study on inventory management demonstrates how segmentation reveals insights into understock and overstock issues, enhancing decision-making in data-intensive domains.