Sentiment analysis from textual data using multiple channels deep learning models

Text sentiment analysis has been of great importance over the last few years. It is being widely used to determine a person’s feelings, opinions and emotions on any topic or for someone. In recent years, convolutional neural networks (CNNs) and long short-term memory (LSTM) have been widely adopted to develop such models. CNN has shown that it can effectively extract local information between consecutive words, but it lacks in extracting contextual semantic information between words. However, LSTM is able to extract some contextual information, where it lacks in extracting local information. To counter such problems, we applied the attention mechanism in our multi-channel CNN with bidirectional LSTM model to give attention to those parts of sentence which have major influence in determining the sentiment of that sentence. Experimental results show that our multi-channel CNN model with bidirectional LSTM and attention mechanism achieved an accuracy of 94.13% which outperforms the traditional CNN, LSTM + CNN and other machine learning algorithms.