A Secure Cloud-Enabled Framework for the Classification of Sentiment Analysis in E-Commerce Using Natural Language Processing with Vision Transformer

Sentiment analysis is very important in the e-commerce business for analyzing customer feedback, product reviews, and preferences on the part of businesses. Processes would reflect the improvement in development, services, and products. This research presents a secure cloud-enabled framework for bidirectional encoder representations from transformers (BERTs) and vision transformer (ViT) for processing huge volumes of text data mined from platforms such as Amazon. The framework uses cloud storage and data management for efficient data handling, scalable architectures, data access, and security while complying with industry regulations. The data collection process is done using millions of customer reviews with the labeling done by sentiment using star ratings. The preprocessed techniques that will clean and prepare the text data for analysis are tokenization and stopword removal. Feature extraction uses BERT to convert their texts into meaning-context aware embeddings that ViT model for further processing of sentiment classification as positive or negative class. Multi-factor authentication (MFA) protects sensitive customer data by putting an extra layer of protection. The cloud infrastructure is scalable, making it possible for the system to offer an upgrade on the increasing data volume. The data management ensures the accuracy and integrity of the data in compliance with standards. With 0.995 accuracy and 0.997 precision, the model enriches sentiment analysis thereby improving product recommendations and bettering decision making. Overall, this would translate into a framework that is highly secured as well as efficient and scalable for e-commerce on sentiment analysis.