Intelligent measuring for a customer satisfaction level inspired by transformation language model

The rapid growth of e-commerce has fundamentally reshaped online consumer behaviour, creating a disconnect between sellers and consumers, and potentially resulting in dissatisfaction. To address this, sentiment analysis emerges as a crucial tool for business decision-makers, providing insights into product and service preferences and a profound understanding of customer sentiments. While conventional machine learning algorithms struggle with intricate patterns, deep learning, especially transformation learning, proves to be a robust solution. Deep learning excels in intricate sentiment classification tasks, yet it demands extensive data, posing challenges for smaller databases. In this paper, we propose a customer satisfaction level framework inspired by the Bidirectional Encoder Representations from the Transformers (BERT) model, The proposed model has the capacity to process bidirectional text contexts and has catalysed a paradigm shift in sentiment analysis. The result demonstrated that our model outperforms other sentiment analysis models.