Enhancing Social Media Sentiment Analysis with a Deep Learning Language Model Ensemble

Abstract

The need for understanding users' actions is significant due to the fast growth of data caused by users' contributions on social media platforms, which is especially pertinent in light of the current epidemic caused by the coronavirus. The scope of the investigated dataset in this research is the thoughts included inside postings relating to the epidemic. It might be difficult to identify the categorization algorithms that are best suited for this sort of information. In this setting, models of deep learning for sentiment analysis have the potential to bring detailed representation capabilities and increased performance in comparison to feature-based methods that are already in use. In this study, we focus on enhancing the performance of sentiment classification by utilizing a specialized deep-learning model. This model combines an improved word embedding method with a long short-term memory (LSTM) network that we construct. Ultimately, our goal is to improve the accuracy of sentiment analysis. In addition, we present an ensemble model that combines our baseline classifier with other state-of-the-art classifiers that are used for conducting sentiment analysis. This model was created by combining our baseline classifier with other state-of-the-art classifiers. It is the goal of this model to be more accurate than any of the other models taken individually. This article made two types of contributions. (1) We provide a robust framework using word embedding and an LSTM network to learn contextual links among words and grasp unheard or unusual words in emergent circumstances like the coronavirus epidemic by detecting suffixes and prefixes from training data. Because it can learn word contexts, this framework can achieve this. This framework is able to do this task as a result of its capacity to learn the contextual relations between words and to learn the contextual connections between words. (2) We propose a hybrid ensemble model for sentiment analysis in order to capture and make use of the major discrepancies that exist among methods that are considered to be state-of-the-art. These discrepancies exist across various approaches to the problem of analyzing people's feelings about things. In a lot of the experiments that we run, we make use of our one-of-a-kind Twitter coronavirus hashtag dataset, as well as public review datasets from Amazon and Yelp. An inquiry based on statistics is carried out for the aim of drawing conclusions, and the findings of this research demonstrate that the performance of these recommended models beats that of other models with respect to the accuracy of categorization