Multimodal Controversy Detection on MMCD rich comments
71.64F1 ScoreAuDisAgent
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| AuDisAgentBackbone LLM=GLM4-9B2026.05 | 71.64 | 72.08 | 71.2 | 71.47 | |
| AgentMCD2026.05 | 69.98 | 70.14 | 70.6 | 70.14 | |
| Zero-shot CoT2026.05 | 69.15 | 69.17 | 69.22 | 69.17 | |
| Self-Reflect2026.05 | 69.13 | 69.26 | 69.58 | 69.26 | |
| Tree of Thoughts (ToT)2026.05 | 68.41 | 68.46 | 68.58 | 68.46 | |
| Plan-and-Solve2026.05 | 68.11 | 68.11 | 68.12 | 68.11 | |
| Standard Prompting2026.05 | 67.91 | 67.93 | 67.98 | 67.93 | |
| Self-Consistency2026.05 | 65.76 | 65.83 | 65.95 | 65.83 | |
| AutoAgents2026.05 | 64.17 | 65.2 | 65.93 | 64.39 | |
| Multi-Agent (Debate)2026.05 | 63.44 | 64.27 | 65.68 | 64.25 | |
| Self-Refine2026.05 | 62.08 | 63.96 | 67.41 | 63.96 | |
| RECITE2026.05 | 61.89 | 63.03 | 64.82 | 63.04 | |
| MAD2026.05 | 61.84 | 61.84 | 61.85 | 61.85 | |
| Cumulative Reasoning (CR)2026.05 | 50.16 | 57.42 | 67.78 | 57.42 |