Improvising E-Commerce Sentiment Analysis with Hybrid VADER-BERT Ensemble Model

With substantial growth in the e-commerce market, the product reviews have taken the key role to support customers in the decision making process. User feedback may contribute to customers making not only the effective decisions but also help the business community to improve in multiple dimensions. Correct sentiment analysis is an important factor in managing product quality. Customers rely on reviews to gauge others' opinions about products. However, existing sentiment analysis models face several challenges. These include difficulties in capturing the complexity and context of language, leading to reduced accuracy. The multimodality of reviews data and the potential imbalanced datasets pose various challenges, consequently resulting in biased classification. This paper proposes a hybrid model combining BERT, SMOTE, and VADER to address these challenges. BERT provides a deep understanding of local and global dependencies in text, SMOTE generates synthetic data to address class imbalance, and VADER provides labels for fine tuning BERT. With our real dataset, the proposed ensemble model achieved 98.1% accuracy with other satisfactory performance measures.