Reproduce “ARBITRA: Robust Document-Level ATT&CK Extraction via Multi-Expert Argumentation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Youngjin Jin, Seungwon Shin
ARBITRA extracts document-level ATT&CK labels from threat reports while limiting unsupported candidates. Three specialists propose labels; unanimous suggestions pass directly, while disputed ones receive arguments for and against, scored by a framework. On TRAM2, its reasoning-fine-tuned version reaches macro precision, recall, and F1 of 0.722, 0.635, and 0.649, outperforming a sentence-level baseline on all three; results do not establish universal superiority.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.