Translations as Additional Contexts for Sentence Classification

This TensorFlow code was used in the experiments of the research paper

Reinald Kim Amplayo, Kyungjae Lee, Jinyoung Yeo, and Seung-won Hwang. Translations as Additional Contexts for Sentence Classification. IJCAI, 2018.

We provided the MR dataset and its translations to 10 other languages in the data folder to try the code.

Additionally, you will need two create three other folders which should contain the following:

• pickles folder: should contain nothing; the code automatically puts the pickled dataset files here
• vectors folder: should contain nothing; the code automatically puts the pickled word vector files here
• pretrained folder: should contain word vectors for all the languages that is used; these can be downloaded here: https://github.com/facebookresearch/fastText/blob/master/pretrained-vectors.md

To run the code, generate the pickle files by executing the following:

python src/CreateDataset.py

To train and test the model, execute the following code:

python src/mcfa.py

To cite the paper/code, please use this BibTex:

@inproceedings{amplayo2017translations,
	Author = {Reinald Kim Amplayo and Kyungjae Lee and Jinyoung Yeo and Seung-won Hwang},
	Booktitle = {IJCAI},
	Location = {Stockholm, Sweden},
	Year = {2018},
	Title = {Translations as Additional Contexts for Sentence Classification},
}

If you have questions, send me an email: rktamplayo at yonsei dot ac dot kr