Downloading the dataset:

• Download the dataset from Kaggle - Toxic Comment Classification - https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge/data

• The root directory should contain the data folder with the dataset from Kaggle.

• Install the modules provided in the requirements.txt

• THE data FOLDER NEEDS TO PUT INTO THE source folder extracted from source.zip.

• For the project.ipynb to be able to display images, Images folder from gitlab needs be downloaded via this link - https://csil-git1.cs.surrey.sfu.ca/krutp/nlpclass-1197-g-lexchunkers/-/archive/master/nlpclass-1197-g-lexchunkers-master.zip?path=project%2FImages

• The images folder should be placed inside the source folder extracted from source.zip

Here's how to run all the three models

For Logistic Regression

• run python3 Log_reg/log_regression.py

For LSTM

• Download crawl-300d-2M.vec and glove.840B.300d.txt. Put them in data folder
• run python3 LSTM/LSTM.py

For TextCNN

• Download crawl-300d-2M.vec.zip and extract it in data folder
• run python3 TextCNN/textCNN.py

NOTE: If something doesn't work just clone the project directory from https://csil-git1.cs.surrey.sfu.ca/krutp/nlpclass-1197-g-lexchunkers/tree/master/project. Word embedding still would have to be downloaded separately. NOTE: Report file project.ipynb contains images so Images folder needs to be downloaded from gitlab

Checking the output files: The output.zip contains all the submission predictions generated by the three models. They should be submitted to Kaggle for evaluation