Social networks evolved as crucial intermediary for information exchange and relationship building, with texts as the major vehicle for information exchange and relationship building. Social media text data enable, in turn, to track and map the “public voice” in near real time. We developed a comprehensive data catalogue (including a database of social media text data, covering multiple countries and topics) and cohesive method library (including a set of NLP techniques, covering topic models and dictionaries) to track and map topics and sentiments in (large-scaled) crowd-sourced social media text data. Further, we demonstrated the applicability of the data catalogue and method library for social science and policy making by assessing the societal perception and affective aspects of COVID-19 pandemic in some ADB member economies. This research offers notable contributions. First, this research provides a knowledge resource to future social science research. Second, this research enables a continuous and granular mapping of topics and sentiments, including underlying linguistic dimensions, in society and, thus, allows to detect trends and anomalies in societal perceptions and responses over time. Finally, this research supports policy makers in developing/evaluating targeted policy interventions by providing a comprehensive perspective on public opinions in near real time at large scale.

JEL codes: D91, C81, C55