Acknowledgements

Wei and Zou 2019, EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks (GitHub)

Setup

Run chmod +x download.sh and ./download.sh

Load data

from DataLoader import DataLoader
dl = DataLoader()
X, y = dl.load_subreddits(subreddits=['leagueoflegends', 'AdviceAnimals'])

Run EDA

dl = DataLoader()
X, y = dl.load_subreddits(subreddits=['leagueoflegends', 'AdviceAnimals'])
dl.export_for_eda(X, y) # exports eda_nlp/data/reddit.txt

Then run python eda_nlp/code/augment.py --input=eda_nlp/data/reddit.txt. Augmented data is outputted to eda_nlp/data/eda_reddit.txt. Refer to EDA repo for details about the file format.

To use the augmented data,

X, y = dl.import_from_eda()