Legal Document Classification on ECtHR B (test)
78.1Macro-avg F1 ScoreLADAN-B
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| LADAN-B2025.12 | 78.1 | — | — | — | 81.2 | 75.4 | — | |
| LADAN-L2025.12 | 77.5 | — | — | — | 80.3 | 75.6 | — | |
| LeSICiN-L2025.12 | 76.9 | — | — | — | 76.8 | 78 | — | |
| AoS modelBackbone Model=all-mpnet-base-v22025.12 | 76.3 | 72.2 | 83.4 | 77.4 | 74.1 | 79.7 | 70.6 | |
| Hier. LegalBERT2025.12 | 74.7 | — | — | 80.4 | — | — | — | |
| LeSICiN-B2025.12 | 74.5 | — | — | — | 76.8 | 73.1 | — | |
| Hier. BERT2025.12 | 73.4 | — | — | 79.7 | — | — | — | |
| Logistic RegressionClass Weighting=balanced2025.12 | 70.1 | 67.5 | 80 | 73.2 | 63.8 | 80.7 | 66.1 | |
| TFIDF-SVM2025.12 | 63.8 | — | — | 73 | — | — | — | |
| LLMPromptBackbone Model=GPT-4o mini2025.12 | 62.5 | 47.3 | 88.6 | 61.7 | 51.1 | 85.8 | 46.3 | |
| LLMPromptBackbone Model=GPT-4o mini, Prompting Strategy=CoT2025.12 | 62.4 | 47.4 | 87.6 | 61.5 | 50.6 | 85.7 | 46.1 | |
| LLMPromptBackbone Model=Mistral-7B-Instruct-v0.3, Prompting Strategy=CoT2025.12 | 61.9 | 48.9 | 84.9 | 62 | 52.6 | 80.4 | 47.5 | |
| ChatGPTPrompting Strategy=zero-shot2025.12 | 55.3 | — | — | 62.8 | — | — | — | |
| LLMPromptBackbone Model=Mistral-7B-Instruct-v0.32025.12 | 54.3 | 50.1 | 64.5 | 56.4 | 56.8 | 61.2 | 41.8 | |
| Logistic Regression2025.12 | 52.8 | 81.8 | 60 | 69.2 | 91.6 | 43.5 | 61.1 |