Author: Linus Lind

Date: 21 March 2024

LICENSE: GNU GPLv3

Toxicity detection with probability based Bag-Of-Words model

metrics.py: methods accuracy, precision, recall, F score

model.py: methods for creating bag-of-words dictionary with probabilities and comparing them to make predictions

preprocessing.py: preprocessing methods. Use process() method to apply everything

run.py: main file to run, use gridsearch=True to optimize threshold value

For plug and play usage, have data set in data folder formatted as csv. sep = ',' with headers: id,text,label

id = unique identifier

text = text

label = 0 or 1, (corresponding to not toxic or toxic)