script.py file is can be uploaded for executing in kaggle kernel........ it performs slot filling only
A Keras implementation of the models described in [Hakkani-Tur et al. (2016)] (https://www.csie.ntu.edu.tw/~yvchen/doc/IS16_MultiJoint.pdf).
This model learns various RNN architectures (RNN, GRU, LSTM, etc.) for joint semantic parsing, where intent prediction and slot filling are performed in a single network model.
- Python
- Numpy
pip install numpy
- Keras and associated Theano or TensorFlow
pip install keras
- H5py
pip install h5py
- Train: word sequences with IOB slot tags and the intent label (data/atis.train.w-intent.iob)
- Test: word sequences with IOB slot tags and the intent label (data/atis.test.w-intent.iob)
You can train and test JointSLU with the following commands:
git clone --recursive https://github.com/yvchen/JointSLU.git
cd JointSLU
You can run a sample tutorial with this command:
bash script/run_sample.sh rnn theano 0 | sh
Then you can see the predicted result in sample/rnn+emb_H-50_O-adam_A-tanh_WR-embedding.test.3
.
To reproduce the work described in the paper. You can run the slot filling only experiment using BLSTM by:
bash script/run_slot.sh blstm theano 0 | sh
You can run the joint frame parsing (intent prediction and slot filling) experiment using BLSTM by:
bash script/run_joint.sh blstm theano 0 | sh
Yun-Nung (Vivian) Chen, [email protected]
Main papers to be cited
@Inproceedings{hakkani-tur2016multi,
author = {Hakkani-Tur, Dilek and Tur, Gokhan and Celikyilmaz, Asli and Chen, Yun-Nung and Gao, Jianfeng and Deng, Li and Wang, Ye-Yi},
title = {Multi-Domain Joint Semantic Frame Parsing using Bi-directional RNN-LSTM},
booktitle = {Proceedings of Interspeech},
year = {2016}
}