[IJCNLP 2017 - Accepted] Multi-tasking deep learning framework that achieves state-of-the-art results in sentiment analysis, topic prediction, and hashtag recommendation.

This project explores emotion classification in social media texts using multiple text representation techniques (TF-IDF, Word2Vec, GloVe) and deep recurrent neural networks (GRU, LSTM, Bi-GRU, Bi-LSTM). The best model achieved an F1-Score of 0.9248, with Word2Vec and Bi-LSTM outperforming other combinations.