Emergency Information Diffusion of COVID 19 using Machine Learning Techniques

Objectives : To find the real time influential user in Twitter to diffuse emergency information globally during COVID 19 using social media. Methods: Twitter API is used to extract the real time COVID related tweets with trending hashtags where 80,000 live tweets are collected to process the data. Machine learning models such as Random Forest, Logistic Regression, Decision Tree, Linear Support Vector Machine, Extreme Gradient Boosting, K-Nearest Neighbor, and Ridge are trained to classify the COVID-related data. K-means clustering groups the communities with respect to zone-wise where the top influencer of the zone is identified using Page Rank influencer score. Findings: Random Forest model performs well with 98% accuracy when compared to other models. A total of 16 zones are identified based on K-mean clustering where the page rank algorithm identifies the top influencer user as 38427065 with 0.5741 influential score and it points to zone 14 Perungudi. The emergency information is disseminated through influencers using the proposed influential node identification technique. Novelty : Our system exhibits the geo graphical representation of the COVID-affected area with respect to user tweets. The emergency information about COVID is disseminated through top zone influencers. Keywords: Social Media, COVID 19, Machine Learning, Influencer, Community