<h1 align=center>Comment Toxicity Classification</h1> This model helps to predict the is a comment/sentence is hateful on various parameters such as toxicity, severe toxicity, obscene, threat, insult and racism.

Test the model here : <a href="https://huggingface.co/spaces/pvcodes/comment_toxicity_classifier">pvcodes/comment_toxicity_classifier</a>

<br>

Working of the Model

• Loading of Data

  The data is fetched from <a href='assets/jigsaw_toxic_challenge/train.csv/train.csv'>csv</a> file, which consist of the comment and attributes such as toxicity, severe toxicity, obscene, threat, insult and racism.

• Preprocessing the comments

  Then the data is tokenized using the TextVectorization method of keras in and embeded

• Creating of <emp>Deep NLP Model</emp>

  For this model we used Keras sequential API with a number of LSTM layers (because they are particulary good while working with sequences)

• The dataset used to train the model is from <a href=https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge>Toxic Comment Classification Challenge</a> from <a href=https://www.kaggle.com>Kaggle</a>.

Note: The compiled data model is available here: <a href='assets/toxicity.h5'>here</a>.

<samp> <p align="center"> ════ ⋆★⋆ ════<br> From <a href="https://github.com/pvcodes/pvcodes">pvcodes</a> </p> </samp>