Binary classification (learner attention detection) on Green Chemistry eye-tracking dataset (test)
80.1AccuracyMajority Class (All ”Attentive”)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Majority Class (All ”Attentive”)Method Category=Baselines2026.04 | 80.1 | 40 | 50 | 44.5 | |
| Proportional Random (P = Freq)Method Category=Baselines2026.04 | 68.1 | 50.1 | 50.1 | 50 | |
| Few-shot (1 Exemplar)Method Category=VLM Strategies, Prompting Strategy=Few-shot, Number of Exemplars=12026.04 | 60.6 | 50.3 | 50.4 | 49.3 | |
| Few-shot (5 Exemplars)Method Category=VLM Strategies, Prompting Strategy=Few-shot, Number of Exemplars=52026.04 | 60.4 | 50.4 | 50.6 | 49.4 | |
| Similarity Matching (Blind)Method Category=VLM Strategies, Prompting Strategy=Similarity Matching2026.04 | 59.7 | 49.4 | 49.2 | 48.3 | |
| Zero-shot (Heuristic CoT)Method Category=VLM Strategies, Prompting Strategy=Zero-shot, Reasoning Method=Heuristic CoT2026.04 | 59.4 | 51.7 | 52.5 | 50 | |
| Uniform Random (P = 0.5)Method Category=Baselines2026.04 | 50 | 50 | 50 | 45 |