Insights from a Large-Scale Discussion on COVID-19 in Collective Intelligence

ABSTRACT

We present a large-scale case study using an agent platform that gathered cross-class public opinions about the COVID-19 situation. Our system extracts valuable insights from users from the perspective of offering suggestions to public health policy makers about what pandemic action is required in Afghanistan. The research goal of this study is to contribute insights about the scale of the issues and their solutions for COVID-19 by facilitating crowd engagement and discussing potential factors to address future policy-making issues. We methodically used social platforms to encourage participation from private citizens and invited health experts and patients into our discussion system in collaboration with Afghanistans Ministry of Public Health. ZOOM webinars were used to host discussions on Facebook and YouTube as a way to promote COVID-related collective social awareness. Our system mediates discussions with an unlimited number of participants and extracts the insights of their opinions in real time. It can also classify discussion insights into four elements and find the relationships among them issues, ideas, merits, and demerits. We present various kinds of statistical information from 2,046 collected opinions of 1,101 participants by both identification of Al insights and human-conducted studies. Our discourse analysis categorized the discussion elements and peoples word-usage frequency in each group. We analyzed the word frequencies of each groups discussion data based on four types of extracted IBIS elements and extracted the most frequently used keywords. In a discourses IBIS analysis, we found that a solution proposed by one class was considered an issue by another class. To the best of our knowledge, this is the first time in Afghanistan that a cross-class dialogue on COVID-19 has been organized using conversational Al as a representative application for public mediation and deliberation.