Advancements in social media sentiment analysis: A study utilizing the PassiveAggressiveClassifier model

Abstract

In recent years, social media sentiment analysis has garnered widespread attention and application within academic circles. By scrutinizing textual data from social media users, insights into users' emotional inclinations and attitudes can be gleaned, thereby furnishing vital references for domains such as corporate marketing and public opinion surveillance. Leveraging the PassiveAggressiveClassifier model as a foundation, this study undertakes sentiment analysis on posts across social media platforms including Instagram, Facebook, and Twitter. Through stages encompassing data preprocessing, model training, and performance evaluation, the model achieves an accuracy rate of 71.43% on the test dataset. Experimental findings indicate commendable proficiency of the model in discerning negative sentiments, albeit encountering challenges in classifying neutral sentiments. Future research endeavors may seek to refine model performance, broaden datasets, and explore novel application scenarios, thereby addressing challenges and meeting evolving demands within the realm of social media sentiment analysis.