| Summary: | Depression is now a common mental health condition that must be treated in people, particularly adolescents. Although the issue of mental illness has received more attention in recent years, many cases of depression remain undiagnosed. Traditionally, the diagnosis process involved face-to-face communication between doctors and patients, with the doctors referring to clinical depression criteria. However, there are few and limited resources and facilities for the early detection and treatment of depression and other mental illnesses. Therefore, to reduce the negative effect on public health, it is important to provide a model to classify depression by using sentiment analysis. Sentiment analysis has recently emerged as a hot topic in classifying the opinions expressed in a piece of writing to determine the information insight and the writer's point of view on a specific issue. This paper investigates the method to classify sentiments in the mental health domain. The classical features and sentiment analysis approach is conducted in this preliminary study to detect depression using a sample of data collected from Twitter. By using a combination of the lexical approach and the Naive Bayes model, the highest accuracy is achieved above 69%. The results indicate a combination of lexical and machine learning outperforms several machine learning approaches for classifying depression in individual tweets. © 2022 IEEE.
|
|---|