When web of science and twitter collaborate: leveraging natural language processing to identify key themes on COVID-19 associated mental health
摘要
During emergencies or events of global concern, various sources of data can be leveraged in order to gain a holistic understanding. Although the academic literature in digital libraries can serve as a central source of information on such topics, the opinions expressed by the public on social media platforms can also offer essential insights. The COVID-19 outbreak had severely impacted the lifestyles and well-being of people worldwide. Consequently, the mental health of the general public and front-line healthcare workers became a topic of major concern that warranted investigations by the scientific communities. The current study provides insights into the various themes discussed in the COVID-19 lature by querying the Web of Science (WoS) core collection database for scholarly works published between November 2019 and January 2021. The study identifies mental health as the key aspect of the pandemic and further explores the state of research centered on pandemic-associated mental health using bibliometric analysis. Subsequently, the study establishes a congruence between the aspects of mental health covered in the literature and the opinions people share on social media platforms such as Twitter. Topic modeling using Word2Vec and K-means clustering with 1.6 million tweets on mental health during the pandemic led to distinct themes that were assigned to themes obtained from the literature. This study has important implications for the advancement of mental health research and the establishment of social media and bibliometric data as complementary sources for topics that concern populations around the world.