Offensive-Language-Detection-using-Backpropagation

Simple neural network to detect offensive language usage

Topic

The dataset for this project is obtained from twitter and contains instances of hate speech, offensive language and neither. Seen as I wanted to try with binary classification I will only use the offensive speech and the non offensive to create two classes for the target. The dataset contains labeled text which I will first process and then train a simple neural network to capture the difference between the two kinds of speechs.

Summary

• Importing libraries
• The dataset
• Converting characters to lower case
• Removing punctuation
• Removing digits
• Tokenizing sentences
• Removing stopwords
• Removing frequent words
• Stemming
• Bag of words
• Train/Test split
• Creating train/test loaders
• Building the classifier
• Training and inference
• Trying out the model
• Conclusion

Libraries

• Pandas
• Numpy
• Torch
• NLTK
• SKlearn

Data source

https://www.kaggle.com/datasets/mrmorj/hate-speech-and-offensive-language-dataset?fbclid=IwAR2kOFFLqhMQrV_K1ThWbpWG21WGPqyn4HOsiLzR3vsXMkQw1rgH9MxGGuc