Exploring Opinionated Sentiments through Web Scraping and Attribute-based Analysis

The majority of customers consult online reviews before choosing which e-commerce services or products to buy. The new aspect of the suggested strategy is the use of opinion mining on customer reviews to assist businesses and organizations in continuously improving their marketing strategies and acquiring a thorough analysis of consumer impressions of their goods and brands. This study employed deep learning convolutional neural networks coupled with convolutional neural networks and long short-term memory (CNN-LSTM) to analyze the sentiment of reviews inside the ecommerce business. The algorithm’s performance on previous iterations of the website was assessed using a web archive to predict how well the algorithm might function in the future. When the words around the pertinent data are categorized under the presumption that the website’s future HTML would seem similar to the current HTML, robust scraping is possible. The algorithm generates a variety of outcomes based on a wide range of factors that can be discovered both in the website’s internal code and in prior versions of the websites. The confusion matrix indicates how well the classification algorithm is executed. In simple terms, we obtained an accuracy of 80.12%. The model performed good job prediction for both classes with high precision and recall scores.