To enhance the precision and comprehensiveness of network security situation assessment, this paper proposes a network security situation assessment algorithm based on the Hadoop platform to address the shortcomings of traditional hierarchical analysis methods in network security situation assessment. The method employs Long Short-Term Memory (LSTM) networks to analyze data from multiple sources in order to extract data features, which are then classified using the Random Forest algorithm. Additionally, the algorithm is deployed in a Hadoop parallel cluster to achieve data fusion. This method effectively improves prediction performance, enhances the quality of network security situation assessment, and provides technical support for the development of network security.