Text-Based Sentiment Analysis Using CNN-GRU Deep Learning Model

Abstract

Sentiment analysis identifies both positive and negative viewpoints from sources like social media, surveys, and reviews by automating text analysis with artificial intelligence (AI). Using data to inform decisions is made easier by this. Deep Learning (DL) has gained a lot of interest in recent years from academia and industry because of its outstanding performance. Convolutional neural networks (CNN) and recurrent neural networks (RNN) are the two deep learning designs that are most frequently utilized. Because they can examine enormous amounts of data, neural networks have the potential to be more accurate in sentiment analysis. Our work utilizes a hybrid model that combines Word2Vec preprocessing with the Gated Recurrent Unit (GRU) from Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN) for sentiment analysis on movie and smartphone reviews. Achieving 99.99% precision, 99.84% recall, and 99.92% accuracy for movie reviews, and 99.08% accuracy, 98.93% precision, 99.40% recall, and 98.93% F1-score for Amazon mobile phone assessments. This paper presents a CNN-GRU-based Word2Vec algorithm for sentiment categorization, which addresses the challenge of evaluating vast amounts of user-generated text data with 99.50% accuracy.