Sentiment Analysis and Emotion Detection using Text Mining in Natural Language Processing

There are many studies being conducted in the field of natural language processing related to sentiment analysis due to the growth of computer science and social media. However, there are many applications of sentiment analysis, and it might be difficult for researchers to review and understand past studies in order to see the trends. Thus, this study proposes trends for sentiment analysis research to assist sentiment analysis researchers. The data sources we obtained from our collections are Google Scholar and Scopus. Therefore, reviews of studies that are using multiple types of modalities have examined change from a particular event and from language itself have been researched more frequently in the last few years. Sentiment analysis is a powerful driver of natural language processing with huge implications for business, social media, healthcare, and disaster response, among others. In this review, we explore the complex landscape of sentiment analysis, its implications, challenges, and recent trends. We cover several important topics, such as the best dataset to use, the best algorithm to use, whether to consider different languages, and various new sentiment tasks. We evaluate the applicability of existing datasets (e.g., IMDB Movie Reviews, Twitter Sentiment Dataset) and deep learning approaches (e.g., BERT) to sentiment analysis. There has been progress in sentiment analysis, however, emerging challenges remain (e.g., sarcasm and irony, ethical challenges, and new domains). We believe sentiment analysis is dynamic in nature, so future research is necessary to capture the complexity of sentiment expression exhibited by humans, as well as to ensure ethical and effective use of sentiment analysis across specializations and languages.