Abstract:
As the novel coronavirus pandemic sweeps the globe and people take to their homes to
avoid getting and spreading the contagion, it makes proper sense that much of the
conversation about this is taking place online. With the rise of Social Media usage, web
surfing and a long period of uncertainty during this Pandemic time, there is a sheer concern
about the mental health and anxiety disorders among people. People are now using the
internet to share information, air their anxieties, and spend time while in quarantine. This
increate rate of online Social Media Use (SMU) opened the possibility to identify some
common traits among people with various mental disorders and anxiety by the large dataset
provided. The moments when those online conversations light up also tell us a lot about
how our feelings around the pandemic are evolving. In recent years, this research area has
started to evolve, but it would be extremely valuable during this crisis period. Although it
is a complex task to perform as mental illness patterns are very complicated, it showed the
light of hope in the past. Previously, adoptive supervised machine learning, such as deep
neural network approaches were used to predict the pattern and level of mental illness; but
they failed due to lack of annotated training data. In this research, we are proposing an
effective machine learning architecture, based on Cluster analysis, Natural Language
Processing (NLP) technique in the analysis of unstructured data extraction from Social
media platforms and couple of psychological screener to classify mental condition of
people.