Depression Classification from Thai Text Twitter Post based on NLP-generated Corpus

Depression is a major mental health problem that affects over 1.5 million people in Thailand. Social media platforms like Twitter have become increasingly popular places for people to express their feelings. This is especially true for people who are struggling with depression. In this study, we aim to classify depressed messages based on Thai Twitter posts. The dataset was collected from 10 depressed users (1407 posts) and 20 non-depressed users (10000 posts). Natural language processing (NLP) was used to tokenize posts and to create a new depression corpus. Then features were derived from Tweet’s time and terms matched with the corpus. We used 5 supervised learning classification algorithms: K-nearest neighbors (KNN), Logistic Regression, Support Vector Machine (SVM), Decision Tree, and Naïve Bayes. The experimental results showed that using the depression corpus with Naïve Bayes yielded the best accuracy of 0.9992.