ConvBiGRU deep learning classifier for sentiment analysis with optimization algorithm

Nowadays sentiment analysis is more familiar in the research field. It includes two methods, which are lexicon-based method and machine learning-based method. The lexicon-based method was used by many researchers and it gained successful accuracy but still, there are some disadvantages like low convergence of words while using multiple domains and it does not provide robust results. To overcome this, we proposed the machine learning technique using a deep learning classifier ConvBiGRU with the two combined feature selection methods: Atomic orbital searching and Least Absolute Shrinkage and the Selection Operator algorithm for the training of the classifier. The training and testing process uses four datasets: Twitter-entity-sentiment-analysis reviews, Twitter Airline reviews, Amazon cell phone reviews, and IMDB movie reviews to predict the classification score such as positive, negative, or neutral. The training and testing of the classifier are measured using the performance metrics such as F1 score, Precision, Recall, Accuracy, AUC and ROC curve. The Twitter-entity-sentiment-analysis review dataset gives the highest of 98%. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.