| 要約: | This study addresses the challenge of identifying key mental health topics within YouTube comments, a resource-rich yet underutilized data source for mental health discourse analysis. The research employed Non-Negative Matrix Factorization (NMF) for topic modeling, coupled with sentiment analysis to map out the landscape of mental health discussions. The methodology pivoted around the data science lifecycle, ensuring systematic data collection, preparation, and modeling. A notable deliverable was a dynamic dashboard visualizing the findings through various graphical representations, including bar graphs, pie chart, donut chart and word clouds. The analysis unearthed prevalent themes and sentiments, offering granular insights into public discourse on mental health. The implications of this study are twofold: informing mental health providers about emerging trends and sentiments for more tailored intervention strategies and assisting content creators and platform managers in shaping more impactful communication strategies around mental health topics. The study's findings underscore the value of machine learning techniques, particularly NMF and sentiment analysis, in gleaning actionable insights from large-scale, unstructured social media datasets, thereby advancing the conversation around mental health awareness and support. © 2024 IEEE.
|
|---|