Table 4.
Document Classification Accuracy for Different Categories for Test 2000 with Maximum Entropy Classification
| Category | Number | Exact match | Partial match |
|---|---|---|---|
| Autophagy | 22 | 59.09% | 68.18% |
| Biogenesis | 132 | 58.33% | 61.36% |
| Cell_adhesion | 133 | 66.17% | 70.68% |
| Cell_cycle | 303 | 45.87% | 68.65% |
| Cell_death | 434 | 75.81% | 79.72% |
| Cell_fusion | 20 | 65.00% | 75.00% |
| Cell_motility | 269 | 71.38% | 74.35% |
| Cell_proliferation | 0 | – | – |
| Cell-cell_signaling | 41 | 73.17% | 92.68% |
| Chemi-mechanical_coupling | 147 | 79.59% | 82.31% |
| Intracellular_protein_traffic | 322 | 68.63% | 72.67% |
| Invasive_growth | 52 | 69.23% | 71.15% |
| Ion_homeostasis | 64 | 79.69% | 81.25% |
| Meiosis | 151 | 77.48% | 82.78% |
| Membrane_fusion | 58 | 48.28% | 53.45% |
| Metabolism | 225 | 67.56% | 74.22% |
| Oncogenesis | 168 | 63.10% | 70.83% |
| Signal_transduction | 302 | 59.93% | 67.55% |
| Sporulation | 49 | 73.47% | 81.63% |
| Stress_response | 253 | 64.82% | 73.52% |
| Transport | 84 | 60.71% | 70.24% |
For each code listed in the first column we list the number of articles for which that code is relevant in the second column. The “Exact Match” column lists the percentage of articles for which the classifier predicts the code listed. Because some abstracts have multiple correct codes, the “Partial Match” column lists the percentage of articles for which the classifier assigned any correct code to the article, even if its is not the listed code.