Analyzing Audience Viewership of OTT, TV, Streaming Platforms, and Social Media Through a Comprehensive Intelligent – Integrated Platform

Abstract

The rise of Over-the-Top (OTT) platforms, streaming services, television, and social media has transformed audience engagement, making sentiment analysis a crucial tool for understanding viewer opinions. Sentiment analysis, combined with machine learning
techniques, enables the classification of audience sentiments into various sentiment categories. This study employs Random Forest (RF), Naïve Bayes (NB), and Support Vector Machine (SVM) to analyze sentiment trends in audience reviews. Data was
collected from multiple sources, including social media discussions, streaming platform reviews, and TV audience feedback, resulting in a corpus of 2,000 text samples. The data was annotated, preprocessed, and classified using the three machine learning models.
Random Forest outperformed the others, achieving 98.5% accuracy, 99.7% recall, and 99.8% precision, demonstrating its robustness in sentiment classification. Naïve Bayes followed with 95.7% accuracy, while SVM achieved 94.8% accuracy. Sentiment
distribution analysis showed that positive sentiment dominated, followed by trust, anticipation, and joy, while negative emotions such as fear, anger, and sadness were less frequent. A word cloud analysis further highlighted key themes related to content quality, storytelling, and viewer engagement. These findings suggest that Random Forest is the most effective model for sentiment classification in audience reviews. This study helps policy makers and stakeholders to analyze and make informed decisions on various crossplatform sentiment trends to enhance audience sentiment analysis across diverse media landscapes.