Evaluating Digital Narratives in Heritage Tourism and Museum: Content Analysis, Toxicity Score, and Sentiment Classification Trough SVM and SMOTE

This research uses the Digital Content Reviews and Analysis Framework to explore the dynamic interplay between digital content, sentiment, and toxicity within the context of heritage tourism at the Sangiran site. The study is driven by the urgency to understand how digital narratives impact public engagement and perception, particularly for heritage sites of global significance. Through a comprehensive analysis, the research evaluates toxicity scores, sentiment classifications, and thematic content across multiple videos related to Sangiran. The toxicity analysis reveals generally low levels of harmful content, with an average score of 0.04717, but identifies occasional peaks, highlighting the potential for negative discourse. Sentiment analysis, conducted using the SVM model enhanced by SMOTE, achieves an accuracy rate of 94.59%, with precision and recall rates of 92.07% and 97.79%, respectively, demonstrating the model's robustness in capturing audience sentiment. Content analysis identifies critical themes, such as human evolution and fossil discoveries, emphasizing the educational value of digital content. The research underscores the importance of curating digital narratives that engage, educate, and foster a positive and respectful discourse. The findings suggest that while digital content successfully educates the audience, managing contentious topics is crucial to maintaining constructive engagement. This study contributes to developing more effective digital strategies for heritage tourism, ensuring the sustainable promotion and preservation of sites like Sangiran while addressing the challenges of online discourse. The research highlights the need for continued exploration of digital content's role in shaping public perceptions of cultural heritage.