Sentiment Analysis on Social Media Instagram of Depression Issues Using Naïve Bayes Method

Abstract

The rapid expansion of the digital era has turned social media, especially Instagram, into a crucial source for examining public sentiment on mental health issues such as depression. Depression, a condition that adversely affects thoughts, behaviours, emotions, and overall mental well-being, is often less apparent than physical health problems, leading to delays in treatment. Low public awareness and societal stigma further aggravate these delays, making sufferers hesitant to seek professional help and more inclined to share their experiences online. This study aims to analyze public sentiment on Instagram concerning depression through the Naïve Bayes (NB) method. It involves developing an application that visualizes analysis reports via bar and pie charts, categorizing public comments on depression as neutral, positive, or negative. Data is sourced from Instagram using keywords related to depression and mental health, with lexicon-based methods for labelling and NB for sentiment classification. The findings show the effectiveness of this method, with the accuracy rate reaching 79%. The dataset consists of 1300 comments collected through web crawling. This evaluation displays the performance results of NB achieving an accuracy of 82.55%. The study aims to offer insights into public opinions on depression, provide datasets for future sentiment analysis research, and assess the NB method's effectiveness in managing complex sentiments on social media, ultimately aiming to improve public understanding and strategies for mental health intervention.