Research
TTPrint: Diverge-then-Converge Verification Achieves High-Precision MITRE ATT&CK Extraction from Threat Reports
Proposes TTPrint, a framework that separates candidate TTP generation from validation using a diverge-then-converge architecture, achieving both high recall and high precision on MITRE ATT&CK technique extraction from cyber threat intelligence reports. Existing LLM approaches hallucinate unsupported techniques; TTPrint's two-stage verification catches these. Applicable to automated threat intelligence pipelines.
Source
↳ Follow the thread