Cyber security threat analysis has become increasingly complex with the rapid growth of digital networks and sophisticated cyber-attacks. Traditional security measures struggle to efficiently detect and mitigate advanced threats. This study explores the use of anomaly detection and graph summarization techniques for efficient cyber security threat analysis. Anomaly detection is leveraged to identify unusual patterns in network traffic, enabling the early detection of potential threats. Graph summarization is utilized to reduce the complexity of network data while preserving essential structural information, facilitating faster and more accurate threat analysis. By combining these approaches, the proposed model enhances the scalability and efficiency of cyber security systems. The study investigates various anomaly detection algorithms, including graph-based and machine learning techniques, and evaluates their effectiveness in detecting advanced persistent threats (APTs) and zero-day attacks. Additionally, graph summarization methods such as clustering and graph coarsening are examined for their impact on processing speed and threat detection accuracy. Experimental results demonstrate significant improvements in threat detection rates and reduction in computational overhead. This research contributes to the development of intelligent cyber security systems capable of real-time threat analysis and proactive defense mechanisms, ensuring enhanced network security in an ever-evolving cyber landscape.