A Hybrid CNN-LSTM Approach for Sentiment Analysis

The significant growth of microblogging in social media has caused companies, media outlets, and analysts to look for efficient methods to extract information about people's perceptions of their products and services. In this article, we propose a deep learning-based approach using a hybrid model that combines a convolutional neural network (CNN) and long short-term memory (LSTM) network to classify people's sentiments on a particular topic. Two datasets are used in the experimentation, the first with two polarities (positive or negative) and the second with three polarities (positive, neutral, or negative). The results of the CNN-LSTM hybrid model in the training and testing stages are acceptable. We calculate several parameters to measure the model's performance, such as accuracy, precision, recall, F1-score, and the area under the ROC curve. The classification results show that the proposed model exceeds 84 and 81% accuracy in the sentiment analysis for each dataset.