Sentiment Analysis of Thai Sentences for Monitoring Mental Health Issues Using Transformer Models

In recent years, mental health issues have become increasingly prevalent. Suicide and self-harm are consequences of severe mental illnesses. Though it can be challenging to identify those suffering from mental health issues, careful observation can help guarantee that those who are struggling with mental health issues receive treatment before their circumstance worsens. Several things, including social competition or familial problems, can lead to mental illness. Because people are using online communication channels more and more, this results in more private time and less social engagement, online technology for communication has additionally influenced social surroundings. This change significantly influences the onset of mental health problems. This research aims to present a sentiment analysis model for Thai sentences using the Transformer Model (ALBERT, RoBERT a, and BERT) to monitor mental health problems using 50,065 emotional sentences expressing negative and positive with web scraping techniques from X (Twitter) and YouTube. The experiments show that ALBERT has the highest processing efficiency when compared to RoBERTa and BERT models with an accuracy percentage of 90%, 85%, and 80%, respectively.