Text classification

This code repository covers multi-category classification and multi-label classification, and each classification problem contains a variety of different algorithms.

Multiclass classification

Gather Data

Use the THUCNews dataset that contains the text of 740,000 news from the thuctc.

Training

python -m text_classification.train_ngram_model

Hyperparameter search

python -m text_classification.tune_ngram_model

Predicting

python -m text_classification.predict_ngram_model

Multi-label classification

Gather Data

Use the THUCNews dataset that contains the text of 740,000 news from the thuctc.

Training

python -m text_classification.train_sequence_model

or train sequence model with fine-tuned pre-trained embeddings

python -m text_classification.train_fine_tuned_sequence_model

Predicting

python -m text_classification.predict_sequence_model