BERT-Powered CVSS v4.0 Scoring Prediction

CVSS v4.0 Prediction Model Using BERT – Project Description

This project aims to build a multi-label classification model that predicts the CVSS (Common Vulnerability Scoring System) Version 4.0 base metric values from textual vulnerability descriptions. The model leverages the power of BERT (Bidirectional Encoder Representations from Transformers) to understand and classify security-related texts.

Key Features:

Input: A vulnerability description in natural language.

Output: Predicted values for the 11 base metrics defined in CVSS v4.0:

Attack Vector (AV)

Attack Complexity (AC)

Attack Requirements (AT)

Privileges Required (PR)

User Interaction (UI)

Vulnerability Confidentiality (VC)

Vulnerability Integrity (VI)

Vulnerability Availability (VA)

System Scope (SC)

System Integrity (SI)

System Availability (SA)

And then you have to calculate severity score of vulnerability description and base vector as well . The dataset is not officially available so you have to create your own synthetic dataset