Unraveling Customer Sentiment in Indian Fashion E-Commerce: A BERT-Based Approach

With the surge of e-commerce growth in India, it has become very necessary for brands to understand the sentiments of customers so that they can improve their offerings and remain in competition. In this study, the transformer-based model, namely BERT is used for conducting sentiment analysis of customer reviews related to famous Indian clothing brands such as Myntra, Ajio and Tata Cliq. Our research utilizes BERT contextual embeddings for the capture of opinions and emotions in customer feedback. In this paper, we will look to prove the efficiency of BERT in sentiment classification regarding Indian fashion e-commerce by analyzing a large corpus of reviews from these popular platforms, as well as compare its accuracy to DistilBERT. The study will identify how this approach extracts meaningful insights from unstructured text data that turn very useful to the brands for understanding customer satisfaction and improvement areas. Sentiment analysis of online cloth industry is one of the rarely explored field and our paper introduces this idea, specifically for the Indian demographic. Our results contribute to the developing literature of sentiment analysis in the Indian e-commerce domain and illustrate some of the many ways in which state-of-the-art natural language processing techniques are put to work in retail.