- Training data: 11314 texts
- Test data: 7532 texts
- Unbalanced datasets (see Report.pdf for details)
- Allocated into 20 categories
- Dataset source: https://scikit-learn.org/0.19/datasets/twenty_newsgroups.html
Model | Accuracy | Precision(*) | Recall(*) | F1-Score(*) |
------------------------------------------------------------
SVM | 0.66 | 0.68 | 0.66 | 0.66 |
BiLSTM | 0.67 | 0.68 | 0.67 | 0.67 |
(*) weighted metrics