Emoji Prediction using LSTM and Naive Bayes

The emojis or emoticons are tiny graphics, generally used during texting through social media platforms. These express a combination of visuals and words simultaneously and provide a different method of interactions. Even after being used quite a significant number of times on social media, their meanings and importance haven't received enough focus and attention via the perspective of Natural Language Processing. In the described project, we perform the analysis of the relationship between emojis and English words and predict the respective emojis for some given text or tweet messages. We have performed multiple methods for word embedding, and our model trained on the Multinomial Naive Bayes and LSTMs. As per the conclusion of our project, the model has shown significant appropriate predictions of emojis on the tweets.