Depression Detection in Online Social Media Users Using Natural Language Processing Techniques

Introduction

The proliferation of Online Social Media (OSM) has enabled individuals to communicate their interests, feelings, emotions, and observations. Such contents are being used for various scientific investigations, including behavioral intentions detection, personalization for recommendations, community detection, etc. In a behavioral and psychological context, such contents may also be used to reveal users’ opinions, preferences, and sentiments towards some events. Moreover, the personality traits of individuals may also be extracted from such content. Such traits may be further used to detect users’ current mental states and disorders and predict their future actions (Back et al., 2010). Among other disorders, major depressive disorder or simply depression is being widely detected by the scholastic Community. Since the “language of depression” can affect others in many ways, it can affect the people who read and follow depressive contents in case the writers have a huge fan following. Therefore, the depression detection problem aims to suggest treatments to the potential users so that extreme reactions may be avoided. As a result, the users may regain the joy of life, and other users may get positivity from such users. Moreover, research has shown promising early screening results as on-time treatment leads to higher working productivity and reduces absenteeism (Abboute et al., 2014).

Figure 1.

Hierarchical segregation of sections

As per World Health Organization (WHO), around 264 million persons suffer from temporary to severe levels of depression around the globe. Mild depression causes the usual mood fluctuations, while the severe condition may also lead to committing suicide. As per estimations by WHO, around 0.8 million persons are committing suicide yearly. Suicide is the second leading cause of death for persons of age 15-29, and most of the users of this age use OSM for sharing their feelings, emotions, and observations (“WHO”, 2021). Individuals share these contents in textual and multimedia forms. Different methods are used for depression detection. For instance, computer vision methods are being used for detecting depression from multimedia data. Moreover, psychiatric questionnaires and tests are used by the practitioners for detecting depression from the patients. Thanks to Natural Language Processing (NLP) tools for enabling the researchers and practitioners to detect the depression from textual data and mine opinion from contents as mentioned earlier shared on OSM. The contents shared in a depressive state of mind may contain unusual words having different polarity levels (Efron & Winget, 2010). For instance, it is researched that the depressive users use more stress-related negative emotions, especially negative adverbs, and adjectives, for example, “gloom,” “lonely,” “lost,” “tired,” “sad,” “miserable.” Moreover, the depressive users use more self-expressing pronouns like “I,” “my,” “me,” “mine,” “myself.” These can be used as features or key points to identify depressive symptoms in OSM users. NLP approaches may classify the depressive and non-depressive contents using lexicon and polarity and other similar information (De Choudhury et al., 2013).

Background

Depression is an emergent mood sickness. More than grief in life's struggles and obstacles. Depression may affect thinking, feeling, and functional activities. It may affect the ability to study, work, eat, sleep and appreciate life. Lack of courage and helpless feeling could be extreme and persistent, with diminutive or no relief. Some health specialists and psychologists define depression as “living in a black hole,” feeling of future trouble, feelings of emptiness, and apathy. It can be different in genders, such as men may feel restless and anger most of the time. No matter how individual experiences this disease, it needs to be treated rightly; otherwise, it can become a severe health condition. Many powerful treatments can be taken to overcome depression to regain the joy of life. But before providing any treatment, we need to identify the person in depression (Higuera, 2021).