CLCD: Cross-Lingual Contradiction Detection

Introduction

In this is repository you will find the data and the code for the experiments reported in the paper A logical-based corpus for cross-lingual evaluation.

Install

To install all the libraries run:

$ bash install.sh

Text Generation

The code to generate the synthetic dataset is in the folder clcd/text_generation. In the same folder you can find a detailed description of all the templates used to generate sentence pairs.

To create all datasets both in English and Portuguese just run:

$ bash gen.sh

the result datasets will be stored in the folder text_gen_output.

The list of all used templates is displayed in the file templates.pdf.

Dataset

All data used in the paper can be found in the folder clcd_datasets/.

Reproducing the experiments

To train the different models you will need to use the file basic_RNN_BERT_train.py. It requires the path for training and testing datasets, it also requires a name to store all the results. For example:

$ python basic_RNN_BERT_train.py clcd_datasets/Portuguese/boolean_coordination_pt_train.csv clcd_datasets/Portuguese/boolean_coordination_pt_test.csv pt_experiments 

Citation

  @misc{clcd2019,
    author = {Felipe Salvatore},
    title = {Cross-Lingual Contradiction Detection},
    year = {2019},
    howpublished = {\url{https://github.com/felipessalvatore/CLCD}},
    note = {commit xxxxxxx}
  }