Sentiment Analysis on GPT-4 with Comparative Models Using Twitter Data

Authors : Mustafa Özel, Özlem Çetinkaya Bozkurt

Pages : 23-33

Doi:10.26650/acin.1418834

View : 62 | Download : 62

Publication Date : 2024-06-28

Article Type : Research Paper

Abstract :Every day, people from all over the world use Twitter to talk about many different topics using hashtags. Since ChatGPT was launched, researchers have been studying how people perceive it in society. This research aims to find out what Turkish Twitter users think about OpenAI’s latest AI model called Generative Pre-trained Transformer 4 (GPT-4). The quantitative data used in this study consist of hashtags on the topic of GPT-4 and involve 2,978 tweets on this topic that were shared on Twitter between March 14-April 9, 2023. The study uses TextBlob sentiment scores to classify the tweets and support vector machines, logistic regression, XGBoost, and random forest algorithms to classify the sentiment of the dataset. The results from the logistic regression, XGBoost, and support vector methods are in close alignment. All parameter findings indicate dependable machine learning, emphasizing the models’ success in classifying tweet sentiment.
Keywords : Sentiment analysis, social media, Twitter, natural language processing