This research investigates the complexities of online discourse by conducting a detailed toxicity and topic analysis of travel vlog content on user-generated platforms. By analyzing 1,503 posts using the Perspective API, the study finds generally low levels of toxicity, with an average toxicity score of 0.06995 and a peak of 0.78207, and similarly low average scores for severe toxicity, identity attack, insult, profanity, and threat (0.00654, 0.01237, 0.03778, 0.06241, and 0.01186, respectively). However, the highest recorded values for these measures—0.45895 for severe toxicity, 0.69287 for identity attack, 0.63084 for insult, 0.81864 for profanity, and 0.51957 for threat—highlight the sporadic presence of harmful content. Advanced clustering techniques, such as HDBScan, k-Means, and Gaussian Mixture models, enable a comprehensive examination of thematic diversity and sentiment distribution within the comments, offering valuable insights into audience engagement and perception. These findings underline the critical need for compelling content moderation and community management strategies to mitigate toxic behaviors and promote a positive digital environment. The study concludes that as digital media evolves, further research into toxicity, thematic content, and user engagement is essential for enhancing theoretical frameworks and practical applications in digital communication.
Copyrights © 2024