code/src/mapper/README.md

CVE-to-TTP Mapper Module

This module implements CVE-to-TTP (Tactics, Techniques, and Procedures) mapping and inference using two complementary approaches:

1. CWE-based Mapping: Uses the existing CWE→CAPEC→TTP mappings generated by the data collection pipeline
2. TIE Inference: Uses the Technique Inference Engine to infer TTPs from CVE descriptions

Prerequisites

1. Run the data collection pipeline first to generate CWE-CAPEC-MITRE mappings:

   python -m src.collectors.main_collector --download
   python -m src.collectors.main_collector --convert
   

2. Install NumPy (optional, for TIE inference):

   pip install numpy
   

3. Download MITRE ATT&CK data (required for TIE inference):

   # Run the setup script to download missing data
   python src/mapper/setup_mapper.py
   
   # Or download manually:
   wget https://raw.githubusercontent.com/mitre-attack/attack-stix-data/master/enterprise-attack/enterprise-attack.json -O src/mapper/tie_models/enterprise-attack.json
   

Usage

1. As a Python Module

from src.mapper.cve_ttp_mapper import CVEtoTTPMapper
from src.mapper.mapper_config import MapperConfig

# Initialize mapper
config = MapperConfig()
mapper = CVEtoTTPMapper(config)

# Map a single CVE
cve_data = {
    "id": "CVE-2021-44228",
    "description": "Apache Log4j2 JNDI injection vulnerability...",
    "cwe_ids": ["CWE-502", "CWE-20"]
}

result = mapper.map_cve_to_ttps(cve_data)
print(f"Found TTPs: {result['ttps']}")

2. Command-Line Interface

# Map a single CVE
python -m src.mapper.mapper_cli single CVE-2021-44228 \
    --description "Remote code execution vulnerability" \
    --cwe CWE-502 --cwe CWE-20

# Process a CVE file
python -m src.mapper.mapper_cli file path/to/cve_data.json \
    --output path/to/mappings.json

# Batch process multiple files
python -m src.mapper.mapper_cli batch path/to/cve_directory \
    --pattern "*.json" \
    --output path/to/output_dir

3. Test the Module

# Run the test suite
python src/mapper/test_simple_mapper.py

4. Batch Processing

# Process CVE data and generate mappings
python src/mapper/run_cve_ttp_mapping.py

How It Works

CWE-based Mapping (Direct)

1. Extracts CWE IDs from CVE data
2. Uses pre-computed CWE→CAPEC→TTP mappings
3. Returns all TTPs associated with the CVE's weaknesses

TIE Inference (ML-based)

1. Extracts keywords from CVE description that indicate techniques
2. Uses pre-trained TIE model to infer related techniques
3. Returns predicted TTPs with confidence scores

Combined Approach

The mapper combines both methods:

• First applies CWE-based mapping for direct associations
• Then uses TIE to infer additional TTPs from the description
• Deduplicates and returns comprehensive results

Output Format

{
  "cve_id": "CVE-2021-44228",
  "ttps": ["T1190", "T1059", "T1203"],
  "methods_used": ["CWE-to-CAPEC-to-TTP", "TIE-inference"],
  "total_ttps_found": 3,
  "details": {
    "cwe_mapping": {
      "method": "CWE-to-CAPEC-to-TTP",
      "mappings": [
        {
          "cwe": "CWE-502",
          "capecs": ["CAPEC-586"],
          "ttps": ["T1203"]
        }
      ]
    },
    "tie_inference": {
      "method": "tie_inference",
      "keyword_ttps": ["T1190"],
      "inferred_ttps": ["T1059"],
      "confidence_scores": {
        "T1059": 0.85
      }
    }
  },
  "mapping_timestamp": "2024-01-01T12:00:00"
}

Configuration

Edit mapper_config.py to customize:

• TIE confidence threshold
• Maximum predictions per CVE
• Input/output paths