Now Streaming Sentiments: A Comparative Analysis of GPT and Gemini in Evaluation of OTT Streaming Show Reviews

There has been a significant rise in the number of users on streaming platforms in the last few years. More and more people consume content on these platforms now. Thus, it becomes imperative to analyze audience and critic reviews of shows on these platforms. This study aims to evaluate the performance of two Large Language Models (LLMs)-GPT-3.5 and Gemini Pro, in analyzing audience and critic reviews of online streaming shows. The research focuses on two key tasks: sentiment analysis of critic reviews and rating prediction based on audience reviews. Additionally, it investigates the impact of few-shot learning on these tasks. This research contributes to our understanding of LLMs' capabilities in sentiment analysis and rating prediction within the context of streaming media reviews, offering insights for both academic research and industry applications in content evaluation and recommendation systems. While few-shot learning improves the performance of both LLMs, by comparing the effectiveness of these models using classification evaluation metrics, we uncover that Gemini Pro performs better in sentiment analysis while GPT -3.5 is more effective for rating prediction.