Spam Classifier

Forking and cloning is welcome. Push changes to get access.

A neural network to filter out valid/legit messages from incoming spams. Might work on deployment soon...

Stuff used: SpamSMS dataset from kaggle Data cleaning and preprocessing with 're' and 'NLTK'

1. Lemmatization since I think it should work better thank Porter Stemming.
2. Bag of Words to convert to vectors as TF-IFD made all the values exteremly small, leading to majority of words being labelled 0 when they should be 1.

might try fixing it and use TF-IFD only An ANN with educated guesses for the most part.