Sentiment Classification of Multidomain Reviews Using Machine Learning Models

Over the past few years, we have witnessed an incredible rise in some of the e-commerce platforms such as Amazon and Flipkart. Platforms like Amazon have changed the way people shop with its vast marketplace. Nowadays, Amazon has become the biggest platform that hosts millions of products and billions of reviews. After the pandemic, when the world almost stopped in-store shopping, nearly 40% of the sales across the e-commerce platforms were held by Amazon. Because of the growing popularity of platforms like Amazon, there has also been a significant rise in product reviews available to consumers. By these reviews, this is going to be easy for the potential buyers and understanding the sentiment expressed in them can provide essential information about the product. For selecting a product, a customer goes through hundreds of reviews before understanding it. But in this world of prospering AI and machine learning, it would be a waste of time to go through all the reviews. This research paper aims to provide a detailed analysis of multidomain product reviews and extract valuable insights for both consumers and sellers, exploring various methodologies and techniques required in this field such as natural language processing techniques, sentiment analysis, and machine learning algorithms to extort important insights through various product reviews. Experimentally, it has been found that SLR and RF algorithms give better classification accuracy of 97.074% and 96.027%, respectively. SLR is having the best TP rate of 0.986 which indicates that retraining g the model is effective. The precision and recall values are highest in the case of SLR which indicates that classification results are accurate.