In this we are correcting the corrupted english text like whr r u going? to proper english text where are u going? using deep learning models.
Dataset is taken from https://www.comp.nus.edu.sg/~nlp/corpora.html which was publically available.
Relevant paper for this problem statement: https://cs224d.stanford.edu/reports/Lewis.pdf
Blog for this problem statement: https://nipun-97agrawal.medium.com/sentence-correction-using-recurrent-neural-network-6321527ee08b
Programming Language used- Python
All the codes are written in Jupyter Notebook of Google Colab
Libraries used in this problem statement
For visulisation Libraries used are:
Here we have used multiple models like
- LSTM and GRU with one hot encoding of characters.
- LSTM and GRU with character tokenization
- LSTM with character tokenization and Bahandanau Attention
- LSTM with fasttext word embedding and Bahadanau Attention