In this play, we build a Vietnamese Text Classifier using VNTC dataset
Experiment results with SVM model and Tfidf, BoW features
| Models | F1 (%) |
|---|---|
| TfidfVectorizer(ngram_range=(1, 2), max_df=0.5) | 92.8 |
| CountVectorizer(ngram_range=(1, 3), max_df=0.7) | 89.3 |
| TfidfVectorizer(max_df=0.8) | 89.0 |
| CountVectorizer(ngram_range=(1, 3) | 88.9 |
| TfidfVectorizer(ngram_range=(1, 3)) | 86.8 |
| CountVectorizer(max_df=0.7) | 85.5 |
Create Development Environment
cd plays/vntc
conda create -n classification python=3.6
pip install -r requirements.txt
Download VNTC dataset
underthesea download-data VNTC
Train a text classifier model
python vntc_train.py
>>> Start training
Dev score: 0.933037037037037
Test score: 0.9267266194191333
>>> Finish training in 116.78 seconds
Your model is saved in tmp/classification_svm_vntc
Predict using trained model
python vntc_predict.py
Text: Huawei có thể không cần Google, nhưng sẽ ra sao nếu thiếu ARM ?
Labels: ['vi_tinh']
Text: Trưởng phòng GD&ĐT xin lỗi vụ học sinh nhận khen thưởng là tờ giấy A4
Labels: ['chinh_tri_xa_hoi']
Optimize hyper-parameters
$ python optimize_hyperparameters.py