Pedagogical strategy recommendation on TeachingStrategiesDataset (test)
55.46PrecisionHybrid-BERT Prob
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Hybrid-BERT Probmechanism=probabilistic weighted sum, weights=BERT (0.5), LPD (0.2), BES (0.3)2026.02 | 55.46 | 53.37 | 48.15 | |
| Hybrid-BERTmechanism=simple majority voting, sources=BM25, embedding similarity, label probabilities, data-augmented BERT2026.02 | 54.7 | 53.89 | 47.52 | |
| BERT-Augmentedbackbone=bert-base-uncased, batch size=8, epochs=10, optimizer=AdamW, learning rate=2e-5, data augmentation=OpenAI GPT API2026.02 | 37.17 | 37.29 | 46.63 | |
| Hybrid-Trad2026.02 | 25.43 | 32.12 | 23.06 | |
| LPD2026.02 | 21.18 | 24.24 | 22.47 | |
| BERT-Basebackbone=bert-base-uncased2026.02 | 19.09 | 31.09 | 20 | |
| BESalpha=0.22026.02 | 15.62 | 31.82 | 20.26 |