Causal Discovery on iTAG-generated corpora
94.7F1GSCITE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SCITEn=32026.04 | 94.7 | 0.304 | 0.12 | |
| LLM-CGn=32026.04 | 90.8 | 0.826 | 0.264 | |
| RuleBayesn=32026.04 | 79.5 | 1.449 | 0.49 | |
| SCITEn=42026.04 | 71.2 | 0.97 | 0.486 | |
| LLM-CGn=42026.04 | 70.5 | 1.452 | 0.611 | |
| SCITEn=52026.04 | 70.2 | 1.527 | 0.895 | |
| LLM-CGn=52026.04 | 69.5 | 2.364 | 1.384 | |
| SCITEn=62026.04 | 69.2 | 1.923 | 1.256 | |
| LLM-CGn=62026.04 | 68.5 | 2.91 | 2.264 | |
| SCITEn=72026.04 | 68.2 | 3.43 | 2.171 | |
| SCITEn=82026.04 | 67.2 | 3.51 | 2.298 | |
| RuleBayesn=42026.04 | 66.1 | 2.902 | 1.706 | |
| RuleBayesn=52026.04 | 59.2 | 4.043 | 2.407 | |
| RuleBayesn=62026.04 | 55.5 | 5.516 | 4.722 | |
| SCITEn=92026.04 | 53.7 | 3.651 | 3.844 | |
| LLM-CGn=72026.04 | 53.2 | 4.11 | 3.015 | |
| LLM-CGn=82026.04 | 52.2 | 5.94 | 4.906 | |
| SCITEn=102026.04 | 51.3 | 5.675 | 3.924 | |
| RuleBayesn=72026.04 | 50.8 | 7.916 | 6.111 | |
| LLM-CGn=92026.04 | 50.7 | 6.02 | 5.896 | |
| RuleBayesn=82026.04 | 45.8 | 7.996 | 9.167 | |
| LLM-CGn=102026.04 | 40.9 | 8.759 | 7.75 | |
| RuleBayesn=92026.04 | 40.7 | 9.278 | 10.341 | |
| RuleBayesn=102026.04 | 35.5 | 9.358 | 10.421 | |
| SCITEn=32026.04 | 0.971 | 0.523 | 0.139 | |
| SCITEn=3, Backbone=DeepSeek-R12026.04 | 0.965 | 0.438 | 0.127 | |
| SCITEn=42026.04 | 0.961 | 0.905 | 0.473 | |
| LLM-CGn=3, Backbone=DeepSeek-R12026.04 | 0.942 | 0.696 | 0.289 | |
| SCITEn=4, Backbone=DeepSeek-R12026.04 | 0.928 | 1.211 | 0.493 | |
| RuleBayesn=3, Backbone=DeepSeek-R12026.04 | 0.791 | 1.548 | 0.442 | |
| SCITEn=52026.04 | 0.767 | 1.582 | 0.714 | |
| SCITEn=62026.04 | 0.757 | 1.983 | 1.584 | |
| SCITEn=5, Backbone=DeepSeek-R12026.04 | 0.749 | 1.803 | 0.831 | |
| SCITEn=72026.04 | 0.747 | 2.469 | 1.664 | |
| SCITEn=6, Backbone=DeepSeek-R12026.04 | 0.695 | 2.249 | 1.413 | |
| SCITEn=7, Backbone=DeepSeek-R12026.04 | 0.685 | 3.494 | 1.898 | |
| LLM-CGn=4, Backbone=DeepSeek-R12026.04 | 0.68 | 2.128 | 0.58 | |
| LLM-CGn=5, Backbone=DeepSeek-R12026.04 | 0.67 | 2.328 | 1.594 | |
| LLM-CGn=6, Backbone=DeepSeek-R12026.04 | 0.62 | 3.858 | 1.622 | |
| LLM-CGn=7, Backbone=DeepSeek-R12026.04 | 0.61 | 3.938 | 2.253 | |
| SCITEn=8, Backbone=DeepSeek-R12026.04 | 0.604 | 3.574 | 2.936 | |
| LLM-CGn=3, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.604 | 0.711 | 0.257 | |
| SCITEn=82026.04 | 0.597 | 4.436 | 1.878 | |
| SCITEn=9, Backbone=DeepSeek-R12026.04 | 0.594 | 4.02 | 3.016 | |
| LLM-CGn=4, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.594 | 0.986 | 0.555 | |
| SCITEn=92026.04 | 0.587 | 4.923 | 4.053 | |
| LLM-CGn=5, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.584 | 1.868 | 1.558 | |
| SCITEn=102026.04 | 0.577 | 7.081 | 4.996 | |
| LLM-CGn=6, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.574 | 3.905 | 2.479 | |
| SCITEn=10, Backbone=DeepSeek-R12026.04 | 0.568 | 4.1 | 4.147 | |
| LLM-CGn=7, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.564 | 3.985 | 4.313 | |
| LLM-CGn=8, Backbone=DeepSeek-R12026.04 | 0.562 | 4.846 | 3.83 | |
| RuleBayesn=32026.04 | 0.549 | 1.347 | 0.52 | |
| RuleBayesn=42026.04 | 0.539 | 2.158 | 1.141 | |
| RuleBayesn=4, Backbone=DeepSeek-R12026.04 | 0.536 | 3.048 | 1.392 | |
| RuleBayesn=5, Backbone=DeepSeek-R12026.04 | 0.526 | 4.016 | 2.397 | |
| RuleBayesn=52026.04 | 0.524 | 3.453 | 2.549 | |
| RuleBayesn=6, Backbone=DeepSeek-R12026.04 | 0.518 | 5.99 | 5.741 | |
| RuleBayesn=7, Backbone=DeepSeek-R12026.04 | 0.501 | 6.07 | 8.028 | |
| RuleBayesn=8, Backbone=DeepSeek-R12026.04 | 0.464 | 8.534 | 8.664 | |
| LLM-CGn=8, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.455 | 5.281 | 5.14 | |
| LLM-CGn=9, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.445 | 8.6 | 5.768 | |
| LLM-CGn=10, model=Qwen3-235B-A22B-Thinking-25072026.04 | 0.435 | 9.519 | 8.223 | |
| LLM-CGn=9, Backbone=DeepSeek-R12026.04 | 0.433 | 5.514 | 5.483 | |
| LLM-CGn=10, Backbone=DeepSeek-R12026.04 | 0.423 | 5.575 | 7.576 | |
| RuleBayesn=9, Backbone=DeepSeek-R12026.04 | 0.391 | 9.551 | 8.891 | |
| RuleBayesn=62026.04 | 0.389 | 5.028 | 5.704 | |
| RuleBayesn=10, Backbone=DeepSeek-R12026.04 | 0.381 | 18.003 | 10.61 | |
| RuleBayesn=72026.04 | 0.379 | 5.108 | 5.784 | |
| RuleBayesn=82026.04 | 0.369 | 6.499 | 9.993 | |
| RuleBayesn=92026.04 | 0.334 | 9.827 | 10.073 | |
| RuleBayesn=102026.04 | 0.324 | 10.186 | 10.153 |