Extracting threat intelligence relations using distant supervision and neural networks
Résumé
Threat intelligence is vital to implementing cyber security. The automated extraction of relations from open-source threat intelligence can greatly reduce the workload of security analysts. However, implementing this feature is hindered by the shortage of labeled training datasets, low accuracy and recall rates of automated models, and limited types of relations that can be extracted.This chapter presents a novel relation extraction framework that employs distant supervision for data annotation and a neural network model for relation extraction. The framework is evaluated by comparing it with several state-of-the-art neural network models. The experimental results demonstrate that it effectively alleviates the data annotation challenges and outperforms the state-of-the-art neural network models.
Domaines
Informatique [cs]Origine | Fichiers produits par l'(les) auteur(s) |
---|