Quantifying the Significance of Cybersecurity Text through Semantic Similarity and Named Entity Recognition

In order to proactively mitigate the risks of cybersecurity, security analysts have to continuously monitor threat information sources. However, the sheer amount of textual information that needs to be processed is overwhelming and requires a great deal of mundane labor. We propose a novel approach to automate this process by analyzing the text document using semantic similarity and Named Entity Recognition (NER) methods. The semantic representation of the given text has been compared with pre-defined "significant" text and, by using a NER model, the assets relevant to the organization are identified. The analysis results then act as features of the linear classifier to generate the significance score. The experimental result shows that the overall system could determine the significance of the text with 78% accuracy.