Sentiment Analysis on YouTube Using Hybrid Model Deep Learning (LSTM)

Abstract

At the same time as information and communication technology (ICT) is expanding at the quickest rate, the amount of online material that is available on social media platforms is growing at an exponential rate. The study of sentiment derived from online evaluations is attracting the interest of researchers from a variety of organizations, including academic institutions, government agencies, and commercial businesses. Within the fields of Machine Learning (ML) and Natural Language Processing (NLP), sentiment analysis has emerged as a prominent area of study interest. In order to get outstanding outcomes in the field of sentiment analysis, Deep Learning (DL) approaches are now being utilized. A BiLSTM + WV model, which stands for a hybrid convolutional neural network and long short-term memory, was presented for the purpose of sentiment analysis in this study. In order to obtain results, our suggested model is now being utilized in conjunction with dropout, max pooling, and batch normalization. The datasets of airline sentiment on Twitter and Airline quality were subjected to experimental analysis. For this purpose, we utilized the Keras word embedding method, which transforms textual data into vectors of numeric values. This method ensures that words that are related to one another have short vector distances between them. In order to evaluate the effectiveness of the model, we computed a number of metrics, including accuracy, precision, recall, and F1-measure, among others. When it comes to sentiment analysis, these parameters for the suggested model are superior to those of the traditional machine learning models. 89.80% accuracy in sentiment analysis is demonstrated by our results analysis, which reveals that the suggested model surpasses competing models.