Deciphering Depression: Linguistic Analysis of Social Media Data

In today’s digital age, where social media and the internet play integral roles in the lives of countless individuals, the ability to discern emotional states holds significant importance, especially for the 300 million people affected by psychological issues. To address this pressing concern, innovative research papers offer a promising avenue for exploration. The initial step in mitigating the impact of such conditions involves discovery. This study focuses on analyzing text-based depression using machine learning techniques. By employing a Linear Support Vector Machine SVM, Decision Tree, Random Forest, Logistic Regression, and Naive Bayes algorithms on the Sentiment140 dataset - 1.6 million tweets, The study aims to identify the common words used by a depressive person and the words used by a Non-depressive person. Various parameters from cumulative distributions are incorporated and the model is dynamically categorized. These features utilized for identification have been extracted from textual content for the Twitter dataset and semantic context. Comparative analysis reveals that the Naïve Bayes detection method yields superior results compared to other machine learning approaches. The findings of this research may catalyze a re-imagining of how individuals’ emotions are predicted, particularly as emotional expressions on online applications such as twitter.