Correlation with Human Judgment on 518 table pairs
0.939Pearson Correlation (r)Claude Opus 4.6
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Claude Opus 4.6Type=LLM2026.03 | 0.939 | 0.89 | 0.804 | |
| Gemini-3-Flash-PreviewType=LLM2026.03 | 0.927 | 0.889 | 0.799 | |
| GPT-5-miniType=LLM2026.03 | 0.888 | 0.827 | 0.739 | |
| DeepSeek-v3.2Type=LLM2026.03 | 0.802 | 0.827 | 0.713 | |
| GriTSConType=Rule-based2026.03 | 0.7 | 0.742 | 0.595 | |
| GriTS-AvgType=Rule-based2026.03 | 0.698 | 0.763 | 0.604 | |
| TEDSType=Rule-based2026.03 | 0.684 | 0.717 | 0.557 | |
| SCORE ContentType=Rule-based2026.03 | 0.641 | 0.654 | 0.522 | |
| SCORE-AvgType=Rule-based2026.03 | 0.637 | 0.684 | 0.539 | |
| GriTSTopType=Rule-based2026.03 | 0.633 | 0.735 | 0.597 | |
| SCORE IndexType=Rule-based2026.03 | 0.558 | 0.681 | 0.558 |