Dynamic Threat Weight of Network Security Communication Based on Multisource Data Analysis

Abstract

While the internet brings convenience to people, there are also many security problems. This study mainly discusses network security communication and ideological security technology based on multisource data analysis. The situation awareness model based on dynamic threat weight is mainly composed of the fusion module, evaluation module, and prediction module. The situation fusion module is mainly based on the fusion algorithm of distance vector optimal classification to fuse multisource data. The situation assessment module is based on the situation assessment method based on a dynamic threat weight to quantitatively evaluate the situation value. The situation understanding part is the basis of situation awareness. Firstly, a large amount of original security data is merged at the data level to obtain a standardized dataset. Then, the relationship between the data is analyzed by the method of correlation analysis, and the security incident information and threat propagation network are obtained. The knowledge training module accurately feeds back the relevant parameters and data, and the knowledge training module adjusts the data accuracy according to the received data accuracy. The main function of the situation fusion module is to process the multisource data and use the fusion algorithm for data fusion. The situation assessment module uses the fused data combined with a specific method for comprehensive assessment and uses visualization technology to show the final results in front of the administrator, which is convenient for more efficient operation. Based on the data fusion module, the situation assessment module judges the current network security status, detects the assessment results through error checking, and feeds back to the knowledge training module. From the trend of the overall network situation, we can see that the probability of the overall network at 4 is the highest, followed by 5. This study makes an effective evaluation of the overall network security.

Detecting Unknown Threat Based on Continuous-Time Dynamic Heterogeneous Graph Network

“…The baseline model is trained by learning behaviours of network entities. Whether a new behaviour is malicious or not is determined by the learned classifier [5][6][7]. In [8], researchers combined metalearning and ensemble learning ideas to enforce anomaly detection.…”