| Classifier | Accuracy |
| Maximum entropy (100 words/category) | 72.12 |
| Naı̈ve Bayes (500 words) | 59.62 |
| Nearest neighbor (5000 words, 50 neighbors) | 61.54 |
[i] For each classification algorithm, we fix the optimal parameters based on the data in Table 3A. The classifier is run with optimal parameters on Test 2001; the accuracy is reported in this table.