Decoding Customer Sentiments in E-Commerce: A Review of Machine Learning and Deep Learning Approaches

Bijendra Singh, Dr.Vijay Kumar

Sentiment Analysis, E-Commerce, Machine Learning, Deep Learning, NLP, Transformer Models, Customer Reviews, Aspect-Based Sentiment Analysis.

E-commerce has been growing exponentially, creating a wave of user-generated content, which includes product reviews, ratings, and social media feedback in the process. Data streams from such analyses will be helpful for businesses to understand better the sentiments of customers, enhance decision-making, and improve customer engagement. Sentiment analysis is one of the critical branches of NLP, where insights are derived from textual data. Some of the classical approaches involve lexicon-based models and Naïve Bayes and Support Vector Machines from the machine learning area. More advanced techniques rely on deep learning approaches, for instance, LSTM, CNN, BERT, GPT-like transformer-based models.Techniques in Sentiment Analysis in E-commerce: A Comparison between Machine Learning and Deep Learning Approaches. This work encompasses major issues, such as sarcasm detection, spam review detection, and aspect-based sentiment analysis. It also includes popularly used datasets, such as Amazon Reviews and Yelp Reviews, and real-world applications, such as personalized recommendation systems and automated customer services. Furthermore, it introduces future research avenues: Explainable AI, multimodal learning, and federated learning for private sentiment analysis. The paper focuses on the review, analysis, methodologies, challenges, and applications in the effort of guiding researchers and industry professionals toward developing more effective sentiment analysis solutions for the e-commerce industry.

Received: 24 Aug 2024, Receive in revised form: 19 Sep 2024, Accepted: 25 Sep 2024, Available online: 30 Sep 2024

10.22161/ijaers.119.7