Sentiment Analysis and Likes Prediction from Social Media Comments: An ML approach
M. Swathi Sree, Mettu Siddhartha, Poli Vamsi Vardhan Reddy, Meena Belwal · 2024
The era of social media connectivity has produced a situation where people have platforms to receive approvals from the whole world whereas they used to be restricted to their local community, and in this work, a detailed analysis of the automated analysis of user-generated content on social media platforms with a primary focus on sentiment analysis of the comments on the posts will be presented. In the current era of social media dominated digital arena with people visiting social media for approval and in need of others’ advice, it will not work to manually analyze the bulk of conversation. To address this challenge of sentimental analysis, we proposeLinear SVC, Logistic Regression, Gradient Boosting Classifier, Ada Boost Classifier, HistGradient Boosting Classifier with lexical analysis. Out of all HistGradient Boosting Classifier is the best model for sentiment analysis, with the highest accuracy mean of 0.935700 and ROC AUC mean of $\mathbf{0. 9 0 5 6 8 5}$, indicating superior performance in classification accuracy and class distinction. It is effective to use this method in analysis of the content of live social activity, and this allows to understand the emotions of the people and their engagement with the content, and this leads to the content optimization and the enhancement of the strategies of audience engagement.