Research on Key Technologies of Network Security Situational Awareness Based on Big Data

This paper delves into the critical technologies for cybersecurity situational awareness in big data environments and proposes a comprehensive framework that integrates feature extraction, assessment and prediction. Firstly, in the feature extraction phase, autoencoders are employed to efficiently extract features from complex network data and identify potential security anomalies. Secondly, an improved Dempster-Shafer (D-S) evidence theory is used to assess cybersecurity threats. By introducing an expectation deviation function for uncertainty variables, the credibility of experts is optimized, and local conflicts among evidence are resolved. Finally, a hybrid model combining uncertain evidence theory with deep learning methods is constructed to predict potential attack scenarios. The model's predictive capabilities and computational efficiency are enhanced through optimization. The research results demonstrate that this framework significantly improves the accuracy and real-time performance of cybersecurity situation assessment, offering substantial social value in addressing cybersecurity challenges in big data environments.