https://jsl5710.github.io/tag/benchmark/BenchmarkHugoBlox Kit (https://hugoblox.com)en-usSun, 01 Feb 2026 00:00:00 +0000https://jsl5710.github.io/media/icon_hu_1b2044c02ce09a43.pnghttps://jsl5710.github.io/tag/benchmark/https://jsl5710.github.io/publication/conference-paper-bluff/Sun, 01 Feb 2026 00:00:00 +0000https://jsl5710.github.io/publication/conference-paper-bluff/<p><strong>BLUFF</strong> is the largest multilingual fake news detection benchmark to date, spanning <strong>79 languages</strong> (20 high-resource &ldquo;big-head&rdquo; + 59 low-resource &ldquo;long-tail&rdquo;) with over <strong>202,000 samples</strong>. The benchmark combines human-written fact-checked content from 130 IFCN-certified organizations with LLM-generated content from 19 diverse models.</p> <p>Key contributions include:</p> <ul> <li><strong>AXL-CoI</strong> (Adversarial Cross-Lingual Agentic Chain-of-Interactions): A multi-agentic framework using 10 fake chains and 8 real chains for controlled multilingual content generation</li> <li><strong>mPURIFY</strong>: A 4-stage quality filtering pipeline with 32 features across 5 dimensions, ensuring dataset integrity through asymmetric evaluation thresholds</li> <li><strong>Bidirectional translation</strong>: English↔X coverage across 70+ languages with 4 prompt variants</li> <li><strong>Comprehensive evaluation</strong>: State-of-the-art detectors suffer up to 25.3% Macro-F1 degradation on low-resource versus high-resource languages</li> </ul> <p>Resources:</p> <ul> <li> </li> <li> </li> <li> </li> <li> </li> </ul>