They believed that the current deep learning-based approaches suffer two limitations; the lack of sentiment information as well as psycholinguistic features and the lack of context information (context-based meaning of words). At last, they used three different models (CNN, GRU and LSTM) to ...