Contradiction Intensity Agreement on RevCI (test)
0.3862Kappa (κ)IMPACT-P
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| IMPACT-PCategory=Our Proposed2026.05 | 0.3862 | 0.6193 | 0.5826 | |
| IMPACT-OACategory=Our Proposed2026.05 | 0.327 | 0.4783 | 0.4421 | |
| Gemini-3 FlashCategory=Single-agent (CoT)2026.05 | 0.3041 | 0.4523 | 0.3799 | |
| CourtEvalCategory=Generic multi-agent2026.05 | 0.286 | 0.41 | 0.349 | |
| ChatEvalCategory=Generic multi-agent2026.05 | 0.281 | 0.402 | 0.341 | |
| MADCategory=Generic multi-agent2026.05 | 0.272 | 0.389 | 0.327 | |
| Self-RefineCategory=Generic multi-agent2026.05 | 0.268 | 0.381 | 0.318 | |
| GPT-5.2Category=Single-agent (CoT)2026.05 | 0.2612 | 0.3679 | 0.3043 | |
| DebateCategory=Generic multi-agent2026.05 | 0.255 | 0.365 | 0.301 | |
| LLaMA-3-8B-InstructCategory=Single-agent (CoT)2026.05 | 0.2488 | 0.3541 | 0.3364 | |
| LLaMA-4 MaverickCategory=Single-agent (CoT)2026.05 | 0.2348 | 0.3315 | 0.2652 | |
| TIDECategory=Our Proposed2026.05 | 0.2202 | 0.3793 | 0.3549 | |
| Qwen-2.5-7B-InstructCategory=Single-agent (CoT)2026.05 | 0.1859 | 0.2854 | 0.2283 |