DeepSentiment: Unlocking Insights from Twitter Data with Advanced Deep Learning Analytics

Abstract

Sentiment analysis learning systems incorporate both textual tweets and reviews. The hybrid model is a combination of the support-vector-machine (SVM), the Long Short-Term Memory (LSTM), and convolution-neural-network (CNN). Each method was tested for reliability and its computational time was measured. When deep-learning (DL) and SVM are merged, it outperforms both approaches on all datasets. DL algorithms have recently proved to offer significant promise in sentiment analysis. Hyperparameter tuning and monitoring for LSTM takes more time but yields better results than for CNN. The integrated model's efficacy varied from job to job, but in every case it outperformed the alternatives. LSTM networks, CNNs, and SVMs are required for DL-SA strategies. All three methods-SVM, LSTM, and CNN-were evaluated for their accuracy and error rates using a hybrid model. The proposed model has been found to be accurate to within 6.76 percentage points of the true value.