Mager et al.45normalized social texts with colloquial errors. Using a sequence-to-sequence neural network, they obtained a 5% improvement in results compared to the previous model. Singh and Singh46corrected spelling errors in Hindi using deep learning. In this process, the misspelled words were ...