<![CDATA[Mental health issues are increasingly gaining attention, with depression being a primary factor linked to high suicide rates caused by psychological disorders. College students are identified as a vulnerable group to depression and anxiety, which can be triggered by various factors. Meanwhile, individuals self-expression on social media, especially on platform X (Twitter), which offers freedom of expression, is considered reflective of one’s mental well-being. This study aims to explore the analysis of college students tweets using a Natural Language Processing (NLP) approach to detect depressive symptoms through linguistic patterns. Data was collected via crawling techniques using keywords such as “depression”, “stress”, and “burnout” resulting in 24,167 tweets from January to March 2025. After data cleaning, 8,308 tweets remained. Sentiment labeling using the Inset Lexicon shows that 68.1% (5,663 tweets) were labeled negative, reflecting college students tendency to use platform X as a medium to express negative emotions. The Random Forest model integrated with TF-IDF feature extraction achieved 87.51% accuracy, demonstrating its capability to address majority class bias (negative) and capture the morphological complexity of informal language. The implications of the research encourage the development of a digital monitoring system for the proactive detection of college college students depression symptoms. The lexicon’s limitations in incorporating informal vocabulary (slang) becomes a recommendation for further research to enhance analysis accuracy.]]>
Copyrights © 2025
