COMPARING SENTIMENT ANALYSIS METHODS: FLIPKART REVIEWS

Sentiment analysis is widely applied to examine opinions, evaluations, attitudes, judgments, and emotions toward a product. In this study, online reviews of the e-commerce portal Flipkart have been analyzed. The dataset contains 189874 rows and 5 columns of product information such as product name, product price, rate, and review and summary of 104 different types of products. Natural Language Processing has been used to carry out the sentiment analysis of online reviews. These techniques are used to conclude the amount of positive and negative reviews received by a product and further identify the opinions of consumers towards the product. In this research, three approaches of sentiment analysis - machine learning, unsupervised lexicon-based analysis and large language models have been used. Machine learning classifiers like logistic regression, support vector machine, decision trees, and eXtreme Gradient Boosting(XGBoost); lexicon-based approaches -Valence Aware Dictionary and sEntiment Reasoner(VADER) lexicon and SentiWordNet; and large language models(LLMs) like Generative pre trained Transformer(GPT)and Bidirectional Encoder Representations from Transformers(BERT) are used. These approaches have been compared based on accuracy, F1-score, recall, precision, and kappa, to determine their effectiveness in sentiment analysis, revealing that machine learning and lexicon-based approaches provide robust performance, while the large language models are computationally intensive, time-consuming, and show comparatively lower accuracy.