<![CDATA[This research employs the CRISP-DM framework to analyze sentiment and toxicity dynamics in tourist vlog reviews thoroughly. The study delves into sentiment classification and toxicity identification nuances by leveraging machine learning algorithms such as k-NN, SVM, NBC, and DT with SMOTE. Utilizing a dataset comprising a substantial number of posts, the analysis reveals varying levels of accuracy across different algorithms. For instance, k-NN and SVM showcase promising accuracy rates of 85.90% and 86.27% in sentiment classification, while NBC and DT with SMOTE yield 72.52% and 71.14%, respectively. Furthermore, the research elucidates the limitations of toxicity analysis, with NBC demonstrating a precision of 64.96% and DT exhibiting lower recall rates. These findings highlight the importance of robust methodologies for understanding sentiment and toxicity dynamics in online content, particularly in tourist vlog reviews.]]>
Copyrights © 2024
