An implementation of intent classification using random forest model on ATIS datasets with scikit-learn.
Datasets downloaded from JointSLU
If you have "models" folder in the root directory, just run:
python run_classifier.py -s "[single_cleaned_sentence_to_classify]"
Else you will have to train a model first with:
python train_classifier.py
This will train a usable model and save it in "models" folder. (Make sure you have placed datasets correctly.)
Given test dataset contains intention "atis_day_name" that train dataset does not contain. This could lead to wrong classification.
I suggest getting a bigger and more even dataset if possible to fix this issue.
2020, net2cn.