AITI: An Automatic Identification Model of Threat Intelligence Based on Convolutional Neural Network

AITI: An Automatic Identification Model of Threat Intelligence Based on Convolutional Neural Network

연구 분야: Safety

학회: ICIAI '20: Proceedings of the 2020 the 4th International Conference on Innovation in Artificial Intelligence

Cyberspace security issues are becoming more and more important, but traditional methods cannot cope with changing cyber-attack methods, which often leads to severe network paralysis and economic losses. Therefore, we propose an automatic identification model of threat intelligence (TI) based on convolutional neural network (CNN) called AITI. At first, the cyber crawler technology is used to obtain unstructured semantic text data from platforms such as security forums and blogs. Next, the semantic text is preprocessed and input into the word embedding model to extract feature vectors. Then, we build a classification model based on CNN and train it on the collected dataset. Finally, the trained CNN model is used to automatically identify TI from unstructured semantic text such as security articles. According to the experiment results, AITI have obtained the accuracy of 90.38% and the F1 Score of 91.16%. The experiments verified that AITI outperforms other models on multiple core metrics. AITI provides a practical solution for TI identification to a certain extent, which can improve the efficiency and accuracy of TI identification.