ARCHIVES

In the age of digitization, Sentiment analysis has been narrowed to three categories: positive, neutral, and negative. While these procedures fine-grained sentiment analysis using aspects extraction yields greater depth and useful knowledge on less explored area like ChatGPT. many machine learning and deep learning techniques used for this like Support vector machine, naïve byes, RNN, decision tree etc. or many of hybrid also from these techniques. Sentiment and aspect-based analysis are commonly used in areas such as e-commerce, hospitality, AI conversational tools—a fast growing and disruptive technology—remain relatively unknown within this context. ChatGPT, being one of its most popular AI conversational agents, generates a large and diversified amount of consumer input, providing a perfect option for this kind of study. The current research uses aspect-based sentiment analysis to analyze labelled ChatGPT consumer ratings from Kaggle, including aspects like adaptability, responsiveness, handling ambiguity, and knowledge accuracy, as well as a "Other" category and sentiment classes which include Very Positive, Positive, Neutral, Negative, and Very Negative. Normalization, noise removal, and tokenization were among the text preprocessing steps. Aspect extraction performed using a rule-based keyword matching strategy. Sentiment classification was performed using the VADER Sentiment Intensity Analyzer. Utilizing an 80-20 train-test split, aspect classification obtained 94.70% accuracy, sentiment classification 86.92% and Overall accuracy is 83.05%, indicating the method's efficacy in evaluating ChatGPT