<![CDATA[This study investigates user engagement within digital environments, explicitly focusing on creative content like music videos, and examines how sentiment and toxicity levels in user interactions influence engagement dynamics. Employing the Digital Content Reviews and Analysis Framework, the study reveals that 95.8% of user interactions exhibit positive or neutral sentiments. In comparison, a notable 4.2% are toxic, reflecting underlying societal tensions and potentially perpetuating negative feedback loops. Analysis of 23,112 posts using the Perspective API shows an average toxicity score of 0.03972, with severe cases reaching up to 0.87787. Scores for severe toxicity, identity attacks, insults, profanity, and threats, although generally low, indicate maximum values of concern, highlighting the need for vigilant monitoring. Sentiment classification results using the VADER model and multiple algorithms demonstrate that the Support Vector Machine (SVM) model achieved the highest accuracy (68.74%) and Area Under Curve (AUC) score (0.686), outperforming other models in distinguishing sentiment. The study's discussion on user engagement suggests that high levels of participation, such as comments, likes, and shares, are indicators of user interest and community identity but are susceptible to being undermined by toxic interactions. These findings emphasize the importance of fostering positive engagement through effective moderation strategies and advanced sentiment analysis tools, ensuring digital platforms remain conducive to constructive dialogue and community building. The research underscores the necessity for sophisticated analytical approaches to navigate the complexities of user behavior in digital spaces, providing critical insights into the interplay between sentiment, engagement, and toxicity in shaping online communities.]]>
Copyrights © 2024
