Speculative language detection on BioScope
96.2AccuracyBigram SVM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Bigram SVM2026.06 | 96.2 | 96.4 | 81.1 | 88.1 | |
| Recursive Neural Tensor Network (RNTN)2026.06 | 95.8 | 90.4 | 86.6 | 88.5 | |
| Unigram SVM2026.06 | 95.7 | 91.1 | 83.5 | 87.1 | |
| Unigram NB2026.06 | 90.3 | 73.8 | 69.4 | 71.5 | |
| Bigram NB2026.06 | 90 | 90.9 | 47.9 | 62.7 | |
| Pattern Matching2026.06 | 80.5 | 46.8 | 98.5 | 63.5 | |
| Paragraph VectorTraining Data=Bioscope + 1000 BioMed articles2026.06 | 76.1 | 34.1 | 36.8 | 35.4 | |
| Paragraph VectorTraining Data=Bioscope + 3000 BioMed articles2026.06 | 73.5 | 32 | 43.3 | 36.8 | |
| Paragraph VectorTraining Data=BioScope only2026.06 | 70.4 | 29.1 | 46.3 | 35.7 |